kochab/src/types/request.rs

use std::ops;
use std::convert::TryInto;
#[cfg(feature = "scgi_srv")]
use std::{
    collections::HashMap,
    convert::TryFrom,
    path::Path,
};
use anyhow::*;
use percent_encoding::percent_decode_str;
use uriparse::URIReference;
#[cfg(feature="user_management")]
use serde::{Serialize, de::DeserializeOwned};
#[cfg(feature = "gemini_srv")]
use ring::digest;

#[cfg(feature="user_management")]
use crate::user_management::{UserManager, User};

pub struct Request {
    uri: URIReference<'static>,
    input: Option<String>,
    certificate: Option<[u8; 32]>,
    trailing_segments: Option<Vec<String>>,
    #[cfg(feature="user_management")]
    manager: UserManager,
    #[cfg(feature = "scgi_srv")]
    headers: HashMap<String, String>,
    #[cfg(feature = "scgi_srv")]
    script_path: Option<String>,
}

impl Request {
    pub fn new(
        #[cfg(feature = "gemini_srv")]
        mut uri: URIReference<'static>,
        #[cfg(feature = "scgi_srv")]
        headers: HashMap<String, String>,
        #[cfg(feature="user_management")]
        manager: UserManager,
    ) -> Result<Self> {
        #[cfg(feature = "scgi_srv")]
        let (mut uri, certificate, script_path) = (
            URIReference::try_from(
                format!(
                    "{}{}",
                    headers.get("PATH_INFO")
                        .context("PATH_INFO header not received from SCGI client")?
                        .as_str(),
                    headers.get("QUERY_STRING")
                        .map(|q| format!("?{}", q))
                        .unwrap_or_else(String::new),
                ).as_str()
            )
                .context("Request URI is invalid")?
                .into_owned(),
            headers.get("TLS_CLIENT_HASH")
                .map(|hsh| {
                    ring::test::from_hex(hsh.as_str())
                        .expect("Received invalid certificate fingerprint from upstream")
                        .try_into()
                        .expect("Received certificate fingerprint of invalid lenght from upstream")
                }),
            headers.get("SCRIPT_PATH")
                .or_else(|| headers.get("SCRIPT_NAME"))
                .cloned()
        );

        uri.normalize();

        let input = match uri.query().filter(|q| !q.is_empty()) {
            None => None,
            Some(query) => {
                let input = percent_decode_str(query.as_str())
                    .decode_utf8()
                    .context("Request URI query contains invalid UTF-8")?
                    .into_owned();
                Some(input)
            }
        };

        Ok(Self {
            uri,
            input,
            #[cfg(feature = "scgi_srv")]
            certificate,
            #[cfg(feature = "gemini_srv")]
            certificate: None,
            trailing_segments: None,
            #[cfg(feature = "scgi_srv")]
            headers,
            #[cfg(feature = "scgi_srv")]
            script_path,
            #[cfg(feature="user_management")]
            manager,
        })
    }

    pub const fn uri(&self) -> &URIReference {
        &self.uri
    }

    #[allow(clippy::missing_const_for_fn)]
    /// All of the path segments following the route to which this request was bound.
    ///
    /// For example, if this handler was bound to the `/api` route, and a request was
    /// received to `/api/v1/endpoint`, then this value would be `["v1", "endpoint"]`.
    /// This should not be confused with [`path_segments()`](Self::path_segments()), which
    /// contains *all* of the segments, not just those trailing the route.
    ///
    /// If the trailing segments have not been set, this method will panic, but this
    /// should only be possible if you are constructing the Request yourself.  Requests
    /// to handlers registered through [`add_route()`](crate::Server::add_route()) will
    /// always have trailing segments set.
    pub fn trailing_segments(&self) -> &Vec<String> {
        self.trailing_segments.as_ref().unwrap()
    }

    /// All of the segments in this path, percent decoded
    ///
    /// For example, for a request to `/api/v1/endpoint`, this would return `["api", "v1",
    /// "endpoint"]`, no matter what route the handler that received this request was
    /// bound to.  This is not to be confused with
    /// [`trailing_segments()`](Self::trailing_segments), which contains only the segments
    /// following the bound route.
    ///
    /// Additionally, unlike `trailing_segments()`, this method percent decodes the path.
    pub fn path_segments(&self) -> Vec<String> {
        self.uri()
            .path()
            .segments()
            .iter()
            .map(|segment| percent_decode_str(segment.as_str()).decode_utf8_lossy().into_owned())
            .collect::<Vec<String>>()
    }

    /// View any input sent by the user in the query string
    ///
    /// Any zero-length input is treated as no input at all, and will be reported as
    /// [`None`].  This is done in order to provide compatibility with the SCGI header
    /// common practice of reporting no query string as a blank input.
    pub fn input(&self) -> Option<&str> {
        self.input.as_deref()
    }

    #[cfg(feature="scgi_srv")]
    /// View any headers sent by the SCGI client
    ///
    /// When an SCGI client delivers a request (e.g. when your gemini server sends a
    /// request to this app), it includes many headers which aren't always included in
    /// the request otherwise.  Bear in mind that **not all SCGI clients send the same
    /// headers**, and these are *never* available when operating in `gemini_srv` mode.
    ///
    /// Some examples of headers mollybrown sets are:
    ///  - `REMOTE_ADDR` (The user's IP address and port)
    ///  - `TLS_CLIENT_SUBJECT_CN` (The CommonName on the user's certificate, when present)
    ///  - `SERVER_NAME` (The host name of the server the request was received on)
    ///  - `SERVER_SOFTWARE` (= "MOLLY_BROWN")
    ///  - `SCRIPT_PATH` (The prefix the script is being served on)
    pub const fn headers(&self) -> &HashMap<String, String> {
        &self.headers
    }

    #[cfg(feature = "gemini_srv")]
    pub (crate) fn set_cert(&mut self, cert: Option<rustls::Certificate>) {
        self.certificate = cert.map(|cert| {
            digest::digest(&digest::SHA256, cert.0.as_ref())
                .as_ref()
                .try_into()
                .expect("SHA256 didn't return 256 bits")
        });
    }

    pub fn set_trailing(&mut self, segments: Vec<String>) {
        self.trailing_segments = Some(segments);
    }

    #[allow(clippy::missing_const_for_fn)]
    /// Get the fingerprint of the certificate the user is connecting with
    pub fn certificate(&self) -> Option<&[u8; 32]> {
        self.certificate.as_ref()
    }

    #[cfg(feature="user_management")]
    /// Attempt to determine the user who sent this request
    ///
    /// May return a variant depending on if the client used a client certificate, and if
    /// they've registered as a user yet.
    pub fn user<UserData>(&self) -> Result<User<UserData>>
    where
        UserData: Serialize + DeserializeOwned
    {
        Ok(self.manager.get_user(self.certificate())?)
    }

    #[cfg(feature="user_management")]
    /// Expose the server's UserManager
    ///
    /// Can be used to query users, or directly access the database
    pub fn user_manager(&self) -> &UserManager {
        &self.manager
    }

    /// Attempt to rewrite an absolute URL against the base path of the SCGI script
    ///
    /// When writing an SCGI script, you cannot assume that your script is mounted on the
    /// base path of "/".  For example, a request to the gemini server for "/app/path"
    /// might be received by your script as "/path" if your script is mounted on "/app/".
    /// In this situation, if you linked to "/", you would be sending users to "/", which
    /// is not handled by your app, instead of "/app/", where you probably intended to
    /// send the user.
    ///
    /// This method attempts to infer where the script is mounted, and rewrite an absolute
    /// url relative to that.  For example, if the application was mounted on "/app/", and
    /// you passed "/path", the result would be "/app/path".
    ///
    /// When running in `gemini_srv` mode, the application is always mounted at the base
    /// path, so this will always return the path unchanged.
    ///
    /// Not all SCGI clients will correctly report the application's path, so this may
    /// fail if unable to infer the correct path.  If this is the case, None will be
    /// returned.  Currently, the SCGI headers checked are:
    ///
    /// * `SCRIPT_PATH` (Used by mollybrown)
    /// * `SCRIPT_NAME` (Used by GLV-1.12556)
    pub fn rewrite_path(&self, path: impl AsRef<str>) -> Option<String> {
        #[cfg(feature = "scgi_srv")] {
            self.script_path.as_ref().map(|base| {
                let base: &Path = base.as_ref();

                // Make path relative
                let mut path_as_path: &Path = path.as_ref().as_ref();
                if path_as_path.is_absolute() {
                    path_as_path = (&path.as_ref()[1..]).as_ref();
                }

                base.join(path_as_path).display().to_string()
            })
        }
        #[cfg(feature = "gemini_srv")] {
            Some(path.to_string())
        }
    }
}

impl ops::Deref for Request {
    type Target = URIReference<'static>;

    fn deref(&self) -> &Self::Target {
        &self.uri
    }
}