webhookey/src/main.rs

#![feature(proc_macro_hygiene, decl_macro)]

use anyhow::{anyhow, bail, Result};
use clap::{crate_authors, crate_version, AppSettings, Parser};
use hmac::{Hmac, Mac, NewMac};
use ipnet::IpNet;
use log::{debug, error, info, trace, warn};
use nom::{
    branch::alt,
    bytes::complete::{tag, take_until},
    combinator::map_res,
    multi::many0,
    sequence::delimited,
    Finish, IResult,
};
use regex::Regex;
use rocket::{
    data::{FromData, ToByteUnit},
    futures::TryFutureExt,
    http::{HeaderMap, Status},
    outcome::Outcome::{self, Failure, Success},
    post, routes,
    tokio::io::AsyncReadExt,
    Data, Request, State,
};
use run_script::ScriptOptions;
use serde::{Deserialize, Serialize};
use sha2::Sha256;
use thiserror::Error;

use std::{
    collections::HashMap,
    fs::File,
    io::BufReader,
    net::{IpAddr, Ipv4Addr, SocketAddr},
};

#[derive(Debug, Error)]
enum WebhookeyError {
    #[error("Could not extract signature from header")]
    InvalidSignature,
    #[error("Unauthorized request from `{0}`")]
    Unauthorized(IpAddr),
    #[error("Unmatched hook from `{0}`")]
    UnmatchedHook(IpAddr),
    #[error("Could not evaluate filter request")]
    InvalidFilter,
    #[error("IO Error")]
    Io(std::io::Error),
    #[error("Serde Error")]
    Serde(serde_json::Error),
    #[error("Regex Error")]
    Regex(regex::Error),
}

#[derive(Debug, Parser)]
enum Command {
    /// Verifies if the configuration can be parsed without errors
    Configtest,
}

#[derive(Debug, Parser)]
#[clap(
    version = crate_version!(),
    author = crate_authors!(", "),
    global_setting = AppSettings::InferSubcommands,
    global_setting = AppSettings::PropagateVersion,
)]
struct Opts {
    /// Provide a path to the configuration file
    #[clap(short, long, value_name = "FILE")]
    config: Option<String>,
    #[clap(subcommand)]
    command: Option<Command>,
}

#[derive(Debug, Deserialize, Serialize)]
#[serde(deny_unknown_fields, untagged)]
enum AddrType {
    IpAddr(IpAddr),
    IpNet(IpNet),
}

impl AddrType {
    fn matches(&self, client_ip: &IpAddr) -> bool {
        match self {
            AddrType::IpAddr(addr) => addr == client_ip,
            AddrType::IpNet(net) => net.contains(client_ip),
        }
    }
}

#[derive(Debug, Deserialize, Serialize)]
#[serde(deny_unknown_fields, rename_all = "lowercase")]
enum IpFilter {
    Allow(Vec<AddrType>),
    Deny(Vec<AddrType>),
}

impl IpFilter {
    fn validate(&self, client_ip: &IpAddr) -> bool {
        match self {
            IpFilter::Allow(list) => list.iter().any(|i| i.matches(client_ip)),
            IpFilter::Deny(list) => !list.iter().any(|i| i.matches(client_ip)),
        }
    }
}

#[derive(Debug, Deserialize, Serialize)]
#[serde(deny_unknown_fields)]
struct Config {
    hooks: HashMap<String, Hook>,
}

#[derive(Debug, Deserialize, Serialize)]
#[serde(deny_unknown_fields)]
struct JsonFilter {
    pointer: String,
    regex: String,
}

impl JsonFilter {
    fn evaluate(&self, data: &serde_json::Value) -> Result<bool, WebhookeyError> {
        trace!(
            "Matching `{}` on `{}` from received json",
            &self.regex,
            &self.pointer,
        );

        let regex = Regex::new(&self.regex).map_err(WebhookeyError::Regex)?;

        if let Some(value) = data.pointer(&self.pointer) {
            let value = get_string(value)?;

            if regex.is_match(&value) {
                debug!("Regex `{}` for `{}` matches", &self.regex, &self.pointer);

                return Ok(true);
            }
        }

        debug!(
            "Regex `{}` for `{}` does not match",
            &self.regex, &self.pointer
        );

        Ok(false)
    }
}

#[derive(Debug, Deserialize, Serialize)]
#[serde(deny_unknown_fields, rename_all = "lowercase")]
enum FilterType {
    And(Vec<FilterType>),
    Or(Vec<FilterType>),
    #[serde(rename = "json")]
    JsonFilter(JsonFilter),
}

impl FilterType {
    fn evaluate(
        &self,
        request: &Request,
        data: &serde_json::Value,
    ) -> Result<bool, WebhookeyError> {
        match self {
            FilterType::And(filters) => {
                let (results, errors): (Vec<_>, Vec<_>) = filters
                    .iter()
                    .map(|filter| filter.evaluate(request, data))
                    .partition(Result::is_ok);

                if errors.is_empty() {
                    Ok(results.iter().all(|r| *r.as_ref().unwrap())) // should never fail
                } else {
                    errors.iter().for_each(|e| {
                        error!("Could not evaluate Filter: {}", e.as_ref().unwrap_err())
                    });

                    Err(WebhookeyError::InvalidFilter)
                }
            }
            FilterType::Or(filters) => {
                let (results, errors): (Vec<_>, Vec<_>) = filters
                    .iter()
                    .map(|filter| filter.evaluate(request, data))
                    .partition(Result::is_ok);

                if errors.is_empty() {
                    Ok(results.iter().any(|r| *r.as_ref().unwrap())) // should never fail
                } else {
                    errors.iter().for_each(|e| {
                        error!("Could not evaluate Filter: {}", e.as_ref().unwrap_err())
                    });

                    Err(WebhookeyError::InvalidFilter)
                }
            }
            FilterType::JsonFilter(filter) => filter.evaluate(data),
        }
    }
}

#[derive(Debug, Deserialize, Serialize)]
#[serde(deny_unknown_fields)]
struct Hook {
    command: String,
    signature: String,
    ip_filter: Option<IpFilter>,
    secrets: Vec<String>,
    filter: FilterType,
}

impl Hook {
    fn get_command(
        &self,
        hook_name: &str,
        request: &Request,
        data: &serde_json::Value,
    ) -> Result<String> {
        trace!("Replacing parameters for command of  hook `{}`", hook_name);

        replace_parameters(&self.command, request.headers(), data)
    }
}

#[derive(Debug)]
struct Hooks {
    inner: HashMap<String, String>,
}

impl Hooks {
    async fn get_commands(request: &Request<'_>, data: Data<'_>) -> Result<Self, WebhookeyError> {
        let mut buffer = Vec::new();
        let size = data
            .open(256_i32.kilobytes())
            .read_to_end(&mut buffer)
            .map_err(WebhookeyError::Io)
            .await?;
        info!("Data of size {} received", size);

        let config = request.guard::<&State<Config>>().await.unwrap(); // should never fail
        let mut valid = false;
        let mut result = HashMap::new();
        let client_ip = &request
            .client_ip()
            .unwrap_or(IpAddr::V4(Ipv4Addr::UNSPECIFIED));

        let hooks = config.hooks.iter().filter(|(name, hook)| {
            if let Some(ip) = &hook.ip_filter {
                accept_ip(name, client_ip, ip)
            } else {
                info!(
                    "Allow hook `{}` from {}, no IP filter was configured",
                    &name, &client_ip
                );
                true
            }
        });

        for (hook_name, hook) in hooks {
            let signature = request
                .headers()
                .get_one(&hook.signature)
                .ok_or(WebhookeyError::InvalidSignature)?;

            let secrets = hook
                .secrets
                .iter()
                .map(|secret| validate_request(secret, signature, &buffer));

            for secret in secrets {
                match secret {
                    Ok(()) => {
                        trace!("Valid signature found for hook `{}`", hook_name);

                        valid = true;

                        let data: serde_json::Value =
                            serde_json::from_slice(&buffer).map_err(WebhookeyError::Serde)?;

                        match hook.filter.evaluate(request, &data) {
                            Ok(true) => match hook.get_command(hook_name, request, &data) {
                                Ok(command) => {
                                    info!("Filter for `{}` matched", &hook_name);
                                    result.insert(hook_name.to_string(), command);
                                    break;
                                }
                                Err(e) => error!("{}", e),
                            },
                            Ok(false) => info!("Filter for `{}` did not match", &hook_name),
                            Err(error) => {
                                error!("Could not match filter for `{}`: {}", &hook_name, error)
                            }
                        }
                    }
                    Err(e) => trace!("Hook `{}` could not validate request: {}", &hook_name, e),
                }
            }
        }

        if !valid {
            return Err(WebhookeyError::Unauthorized(*client_ip));
        }

        Ok(Hooks { inner: result })
    }
}

#[rocket::async_trait]
impl<'r> FromData<'r> for Hooks {
    type Error = WebhookeyError;

    async fn from_data(
        request: &'r Request<'_>,
        data: Data<'r>,
    ) -> Outcome<Self, (Status, Self::Error), Data<'r>> {
        match Hooks::get_commands(request, data).await {
            Ok(hooks) => {
                if hooks.inner.is_empty() {
                    let client_ip = &request
                        .client_ip()
                        .unwrap_or(IpAddr::V4(Ipv4Addr::UNSPECIFIED));

                    warn!("Unmatched hook from {}", &client_ip);
                    return Failure((Status::NotFound, WebhookeyError::UnmatchedHook(*client_ip)));
                }

                Success(hooks)
            }
            Err(WebhookeyError::Unauthorized(e)) => {
                error!("{}", WebhookeyError::Unauthorized(e));
                Failure((Status::Unauthorized, WebhookeyError::Unauthorized(e)))
            }
            Err(e) => {
                error!("{}", e);
                Failure((Status::BadRequest, e))
            }
        }
    }
}

fn accept_ip(hook_name: &str, client_ip: &IpAddr, ip: &IpFilter) -> bool {
    if ip.validate(client_ip) {
        info!("Allow hook `{}` from {}", &hook_name, &client_ip);
        return true;
    }

    warn!("Deny hook `{}` from {}", &hook_name, &client_ip);
    false
}

fn validate_request(secret: &str, signature: &str, data: &[u8]) -> Result<()> {
    let mut mac = Hmac::<Sha256>::new_varkey(secret.as_bytes())
        .map_err(|e| anyhow!("Could not create hasher with secret: {}", e))?;
    mac.update(data);
    let raw_signature = hex::decode(signature.as_bytes())?;
    mac.verify(&raw_signature).map_err(|e| anyhow!("{}", e))
}

fn get_header_field<'a>(headers: &'a HeaderMap, param: &[&str]) -> Result<&'a str> {
    headers
        .get_one(
            param
                .get(1)
                .ok_or_else(|| anyhow!("Missing parameter for `header` expression"))?,
        )
        .ok_or_else(|| anyhow!("Could not extract event parameter from header"))
}

fn get_value_from_pointer<'a>(data: &'a serde_json::Value, pointer: &'a str) -> Result<&'a str> {
    let value = data
        .pointer(pointer)
        .ok_or_else(|| anyhow!("Could not get field from pointer {}", pointer))?;

    value
        .as_str()
        .ok_or_else(|| anyhow!("Could not convert value `{}` to string", value))
}

fn replace_parameters(
    input: &str,
    headers: &HeaderMap,
    data: &serde_json::Value,
) -> Result<String> {
    let parse: IResult<&str, Vec<&str>> = many0(alt((
        map_res(
            delimited(tag("{{"), take_until("}}"), tag("}}")),
            |param: &str| {
                let expr = param.trim().split(' ').collect::<Vec<&str>>();

                match expr.get(0) {
                    Some(&"header") => get_header_field(headers, &expr),
                    Some(pointer) => get_value_from_pointer(data, pointer),
                    None => bail!("Missing expression in `{}`", input),
                }
            },
        ),
        take_until("{{"),
    )))(input);

    let (last, mut result) = parse
        .finish()
        .map_err(|e| anyhow!("Could not parse command: {}", e))?;
    result.push(last);

    Ok(result.join(""))
}

fn get_string(value: &serde_json::Value) -> Result<String, WebhookeyError> {
    match &value {
        serde_json::Value::Bool(bool) => Ok(bool.to_string()),
        serde_json::Value::Number(number) => Ok(number.to_string()),
        serde_json::Value::String(string) => Ok(string.as_str().to_string()),
        x => {
            error!("Could not get string from: {:?}", x);
            unimplemented!()
        }
    }
}

#[post("/", format = "json", data = "<hooks>")]
async fn receive_hook<'a>(address: SocketAddr, hooks: Hooks) -> Status {
    info!("Post request received from: {}", address);

    hooks.inner.iter().for_each(|(name, command)| {
        info!("Execute `{}` from hook `{}`", &command, &name);

        match run_script::run(command, &vec![], &ScriptOptions::new()) {
            Ok((status, stdout, stderr)) => {
                info!("Command `{}` exited with return code: {}", &command, status);
                trace!("Output of command `{}` on stdout: {:?}", &command, &stdout);
                debug!("Output of command `{}` on stderr: {:?}", &command, &stderr);
            }
            Err(e) => {
                error!("Execution of `{}` failed: {}", &command, e);
            }
        }
    });

    Status::Ok
}

fn get_config() -> Result<File> {
    // Look for config in CWD..
    if let Ok(config) = File::open("config.yml") {
        info!("Loading configuration from `./config.yml`");

        return Ok(config);
    }

    // ..look for user path config..
    if let Some(mut path) = dirs::config_dir() {
        path.push("webhookey/config.yml");

        if let Ok(config) = File::open(&path) {
            info!(
                "Loading configuration from `{}`",
                path.to_str().unwrap_or("<path unprintable>"),
            );

            return Ok(config);
        }
    }

    // ..look for systemwide config..
    if let Ok(config) = File::open("/etc/webhookey/config.yml") {
        info!("Loading configuration from `/etc/webhookey/config.yml`");

        return Ok(config);
    }

    // ..you had your chance.
    bail!("No configuration file found.");
}

#[rocket::main]
async fn main() -> Result<()> {
    env_logger::init();

    let cli: Opts = Opts::parse();

    let config: Config = match cli.config {
        Some(config) => serde_yaml::from_reader(BufReader::new(File::open(config)?))?,
        _ => serde_yaml::from_reader(BufReader::new(get_config()?))?,
    };

    trace!("Parsed configuration:\n{}", serde_yaml::to_string(&config)?);

    if cli.command.is_some() {
        debug!("Configtest succeded.");
        println!("Config is OK");
        return Ok(());
    }

    rocket::build()
        .mount("/", routes![receive_hook])
        .manage(config)
        .launch()
        .await?;

    Ok(())
}

#[cfg(test)]
mod tests {
    use super::*;
    use rocket::{
        http::{ContentType, Header},
        local::asynchronous::Client,
    };
    use serde_json::json;

    #[rocket::async_test]
    async fn secret() {
        let mut hooks = HashMap::new();

        hooks.insert(
            "test_hook".to_string(),
            Hook {
                command: "".to_string(),
                signature: "X-Gitea-Signature".to_string(),
                ip_filter: None,
                secrets: vec!["valid".to_string()],
                filter: FilterType::JsonFilter(JsonFilter {
                    pointer: "*".to_string(),
                    regex: "*".to_string(),
                }),
            },
        );

        let config = Config { hooks: hooks };

        let rocket = rocket::build()
            .mount("/", routes![receive_hook])
            .manage(config);

        let client = Client::tracked(rocket).await.unwrap();
        let response = client
            .post("/")
            .header(Header::new(
                "X-Gitea-Signature",
                "28175a0035f637f3cbb85afee9f9d319631580e7621cf790cd16ca063a2f820e",
            ))
            .header(ContentType::JSON)
            .remote("127.0.0.1:8000".parse().unwrap())
            .body(&serde_json::to_string(&json!({ "foo": "bar" })).unwrap())
            .dispatch();

        assert_eq!(response.await.status(), Status::NotFound);

        let response = client
            .post("/")
            .header(Header::new("X-Gitea-Signature", "beef"))
            .header(ContentType::JSON)
            .remote("127.0.0.1:8000".parse().unwrap())
            .body(&serde_json::to_string(&json!({ "foo": "bar" })).unwrap())
            .dispatch();

        assert_eq!(response.await.status(), Status::Unauthorized);

        let response = client
            .post("/")
            .header(Header::new(
                "X-Gitea-Signature",
                "c5c315d76318362ec129ca629b50b626bba09ad3d7ba4cc0f4c0afe4a90537a0",
            ))
            .header(ContentType::JSON)
            .remote("127.0.0.1:8000".parse().unwrap())
            .body(r#"{ "not_secret": "invalid" "#)
            .dispatch();

        assert_eq!(response.await.status(), Status::BadRequest);

        let response = client
            .post("/")
            .header(Header::new("X-Gitea-Signature", "foobar"))
            .header(ContentType::JSON)
            .remote("127.0.0.1:8000".parse().unwrap())
            .dispatch();

        assert_eq!(response.await.status(), Status::Unauthorized);
    }

    #[test]
    fn parse_command() {
        let mut headers = HeaderMap::new();
        headers.add_raw("X-Gitea-Event", "something");

        assert_eq!(
            replace_parameters("command", &headers, &serde_json::Value::Null).unwrap(),
            "command"
        );

        assert_eq!(
            replace_parameters("    command", &headers, &serde_json::Value::Null).unwrap(),
            "    command"
        );

        assert_eq!(
            replace_parameters("command    ", &headers, &serde_json::Value::Null).unwrap(),
            "command    "
        );

        assert_eq!(
            replace_parameters("    command    ", &headers, &serde_json::Value::Null).unwrap(),
            "    command    "
        );

        assert_eq!(
            replace_parameters("command command    ", &headers, &serde_json::Value::Null).unwrap(),
            "command command    "
        );

        assert_eq!(
            replace_parameters("{{ /foo }}   command", &headers, &json!({ "foo": "bar" })).unwrap(),
            "bar   command"
        );

        assert_eq!(
            replace_parameters(
                "   command  {{ /foo }}  ",
                &headers,
                &json!({ "foo": "bar" })
            )
            .unwrap(),
            "   command  bar  "
        );

        assert_eq!(
            replace_parameters(
                "{{ /foo }}   command{{/field1/foo}}",
                &headers,
                &json!({ "foo": "bar", "field1": { "foo": "baz" } })
            )
            .unwrap(),
            "bar   commandbaz"
        );

        assert_eq!(
            replace_parameters(
                "   command  {{ /foo }}  ",
                &headers,
                &json!({ "foo": "bar" })
            )
            .unwrap(),
            "   command  bar  "
        );

        assert_eq!(
            replace_parameters(
                "   {{ /field1/foo }}   command",
                &headers,
                &json!({ "field1": { "foo": "bar" } })
            )
            .unwrap(),
            "   bar   command"
        );

        assert_eq!(
            replace_parameters(
                "   {{ header X-Gitea-Event }}   command",
                &headers,
                &json!({ "field1": { "foo": "bar" } })
            )
            .unwrap(),
            "   something   command"
        );

        assert_eq!(
            replace_parameters(
                "   {{ header X-Gitea-Event }} {{ /field1/foo }}  command",
                &headers,
                &json!({ "field1": { "foo": "bar" } })
            )
            .unwrap(),
            "   something bar  command"
        );
    }

    #[rocket::async_test]
    async fn parse_command_request() {
        let mut hooks = HashMap::new();

        hooks.insert(
            "test_hook".to_string(),
            Hook {
                command:
                    "/usr/bin/echo {{ /repository/full_name }} --foo {{ /pull_request/base/ref }}"
                        .to_string(),
                signature: "X-Gitea-Signature".to_string(),
                ip_filter: None,
                secrets: vec!["valid".to_string()],
                filter: FilterType::JsonFilter(JsonFilter {
                    pointer: "/foo".to_string(),
                    regex: "bar".to_string(),
                }),
            },
        );

        hooks.insert(
            "test_hook".to_string(),
            Hook {
                command: "/usr/bin/echo {{ /repository/full_name }} {{ /pull_request/base/ref }}"
                    .to_string(),
                signature: "X-Gitea-Signature".to_string(),
                ip_filter: None,
                secrets: vec!["valid".to_string()],
                filter: FilterType::JsonFilter(JsonFilter {
                    pointer: "/foo".to_string(),
                    regex: "bar".to_string(),
                }),
            },
        );

        let config = Config {
            // default: None,
            hooks: hooks,
        };

        let rocket = rocket::build()
            .mount("/", routes![receive_hook])
            .manage(config);

        let client = Client::tracked(rocket).await.unwrap();

        let response = client
            .post("/")
            .header(Header::new(
                "X-Gitea-Signature",
                "693b733871ecb684651a813c82936df683c9e4a816581f385353e06170545400",
            ))
            .header(ContentType::JSON)
            .remote("127.0.0.1:8000".parse().unwrap())
            .body(
                &serde_json::to_string(&json!({
                    "foo": "bar",
                    "repository": {
                        "full_name": "keith"
                    },
                    "pull_request": {
                        "base": {
                            "ref": "main"
                        }
                    }
                }))
                .unwrap(),
            )
            .dispatch();

        assert_eq!(response.await.status(), Status::Ok);
    }

    #[rocket::async_test]
    async fn parse_invalid_command_request() {
        let mut hooks = HashMap::new();

        hooks.insert(
            "test_hook".to_string(),
            Hook {
                command: "/usr/bin/echo {{ /repository/full }} {{ /pull_request/base/ref }}"
                    .to_string(),
                signature: "X-Gitea-Signature".to_string(),
                ip_filter: None,
                secrets: vec!["valid".to_string()],
                filter: FilterType::JsonFilter(JsonFilter {
                    pointer: "/foo".to_string(),
                    regex: "bar".to_string(),
                }),
            },
        );

        let config = Config {
            // default: None,
            hooks: hooks,
        };

        let rocket = rocket::build()
            .mount("/", routes![receive_hook])
            .manage(config);

        let client = Client::tracked(rocket).await.unwrap();

        let response = client
            .post("/")
            .header(Header::new(
                "X-Gitea-Signature",
                "693b733871ecb684651a813c82936df683c9e4a816581f385353e06170545400",
            ))
            .header(ContentType::JSON)
            .remote("127.0.0.1:8000".parse().unwrap())
            .body(
                &serde_json::to_string(&json!({
                    "foo": "bar",
                    "repository": {
                        "full_name": "keith"
                    },
                    "pull_request": {
                        "base": {
                            "ref": "main"
                        }
                    }
                }))
                .unwrap(),
            )
            .dispatch();

        assert_eq!(response.await.status(), Status::NotFound);
    }
}