ordinary-config 0.11.1

#![cfg_attr(docsrs, feature(doc_cfg))]
#![doc = include_str!("../README.md")]
#![doc = include_str!("../docs/app-config-reference.md")]
#![doc = include_str!("../docs/host-config-reference.md")]
#![warn(clippy::all, clippy::pedantic)]
#![allow(clippy::missing_errors_doc)]

// Copyright (C) 2026 The Ordinary Authors.
//
// SPDX-License-Identifier: BSD-3-Clause

#[cfg(feature = "docs")]
pub mod jsonschema;
#[cfg(feature = "docs")]
pub use schemars;

mod app;
pub mod auth;
mod host;
mod http;
mod validate;

pub use app::*;
pub use auth::*;
pub use host::*;
pub use http::*;

pub use crate::validate::DOMAIN_REGEX;
use crate::validate::validate;

use anyhow::bail;
use hashbrown::{HashMap, HashSet};
use serde::{Deserialize, Serialize};
use smallvec::smallvec;
use std::collections::BTreeMap;
use std::fmt::Write;
use std::path::Path;
use std::process::Command;
use std::{env, fs};
use tracing::instrument;

#[cfg_attr(feature = "utoipa", derive(utoipa::ToSchema))]
#[cfg_attr(feature = "docs", derive(schemars::JsonSchema))]
#[derive(Deserialize, Serialize, Debug, Clone)]
pub struct ClientLoggingConfig {
    /// bottom end of the delayed delivery range (seconds)
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    min_delay: Option<u32>,
    /// top end of delayed delivery range (seconds)
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    max_delay: Option<u32>,
    /// max number of events to be buffered on the client
    /// prior to flush.
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    max_buffer: Option<u16>,
    /// sets the max number of events in a given request.
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    max_batch: Option<u16>,
}

#[cfg_attr(feature = "utoipa", derive(utoipa::ToSchema))]
#[cfg_attr(feature = "docs", derive(schemars::JsonSchema))]
#[derive(Deserialize, Serialize, Debug, Clone)]
pub enum RedactedHashAlg {
    Blake2,
    Blake3,
}

#[cfg_attr(feature = "utoipa", derive(utoipa::ToSchema))]
#[cfg_attr(feature = "docs", derive(schemars::JsonSchema))]
#[derive(Deserialize, Serialize, Debug, Clone)]
pub struct ServerLoggingConfig {
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub ips: Option<bool>,
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub headers: Option<bool>,
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub credentials: Option<RedactedHashAlg>,
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub timing: Option<bool>,
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub sizes: Option<bool>,
}

#[cfg_attr(feature = "utoipa", derive(utoipa::ToSchema))]
#[cfg_attr(feature = "docs", derive(schemars::JsonSchema))]
#[derive(Deserialize, Serialize, Debug, Clone)]
pub struct LoggingConfig {
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub client: Option<ClientLoggingConfig>,
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub server: Option<ServerLoggingConfig>,
}

/// Compression algorithms
#[cfg_attr(feature = "utoipa", derive(utoipa::ToSchema))]
#[cfg_attr(feature = "docs", derive(schemars::JsonSchema))]
#[derive(Deserialize, Serialize, Debug, Clone, PartialEq)]
pub enum CompressionAlgorithm {
    Uncompressed,
    Gzip,
    Zstd { level: u8 },
    Brotli,
    Deflate,
    All,
}

impl CompressionAlgorithm {
    #[must_use]
    pub fn as_u8(&self) -> u8 {
        match self {
            Self::Uncompressed => 0,
            Self::Gzip => 1,
            Self::Zstd { level: _ } => 2,
            Self::Brotli => 3,
            Self::Deflate => 4,
            Self::All => 255,
        }
    }

    #[must_use]
    pub fn from_u8(val: u8, lvl: Option<u8>) -> Self {
        match val {
            0 => Self::Uncompressed,
            1 => Self::Gzip,
            2 => Self::Zstd {
                level: lvl.unwrap_or(17),
            },
            3 => Self::Brotli,
            4 => Self::Deflate,
            _ => Self::All,
        }
    }

    #[must_use]
    pub fn as_char(&self) -> char {
        match self {
            Self::Uncompressed => '0',
            Self::Gzip => '1',
            Self::Zstd { level: _ } => '2',
            Self::Brotli => '3',
            Self::Deflate => '4',
            Self::All => 'A',
        }
    }

    #[must_use]
    pub fn as_str(&self) -> &'static str {
        match self {
            Self::Uncompressed => "uncompressed",
            Self::Gzip => "gzip",
            Self::Zstd { level: _ } => "zstd",
            Self::Brotli => "br",
            Self::Deflate => "deflate",
            Self::All => "all",
        }
    }
}

#[cfg_attr(feature = "utoipa", derive(utoipa::ToSchema))]
#[cfg_attr(feature = "docs", derive(schemars::JsonSchema))]
#[derive(Deserialize, Serialize, Debug, Clone, Default)]
pub struct ErrorConfig {
    /// Refers to the asset by path.
    ///
    /// Returns as the fallback when a route is missing.
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub asset: Option<String>,
}

#[cfg_attr(feature = "utoipa", derive(utoipa::ToSchema))]
#[cfg_attr(feature = "docs", derive(schemars::JsonSchema))]
#[derive(Deserialize, Serialize, Debug, Clone)]
pub enum RuntimeMode {
    /// Application will run on the shared multithreaded
    /// tokio runtime.
    Shared,
    /// Application will run on a separate thread with its
    /// own single-threaded tokio runtime.
    SingleThreaded,
    /// Application will run on a separate thread with its
    /// own multithreaded tokio runtime.
    MultiThreaded,
}

#[cfg_attr(feature = "utoipa", derive(utoipa::ToSchema))]
#[cfg_attr(feature = "docs", derive(schemars::JsonSchema))]
#[derive(Deserialize, Serialize, Debug, Clone)]
pub struct LifecycleBeforeAfterScripts {
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub before: Option<Vec<Vec<String>>>,
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub after: Option<Vec<Vec<String>>>,
}

#[cfg_attr(feature = "utoipa", derive(utoipa::ToSchema))]
#[cfg_attr(feature = "docs", derive(schemars::JsonSchema))]
#[derive(Deserialize, Serialize, Debug, Clone)]
pub struct TopLevelLifecycle {
    /// run before every lifecycle operation
    pub before_all: Option<Vec<Vec<String>>>,

    /// configure build lifecycle hooks
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub build: Option<LifecycleBeforeAfterScripts>,
}

/// Config definition for an Ordinary Application
#[cfg_attr(feature = "utoipa", derive(utoipa::ToSchema))]
#[cfg_attr(feature = "docs", derive(schemars::JsonSchema))]
#[derive(Deserialize, Serialize, Debug, Clone, Default)]
pub struct OrdinaryConfig {
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub lifecycle: Option<TopLevelLifecycle>,

    /// Domain name for the application to be run from the
    /// deployment environment.
    pub domain: String,

    /// Version of the site build.
    pub version: String,

    /// additional domains with an ALIAS record
    /// pointing at the primary `OrdinaryConfig::domain`.
    ///
    /// add a TXT record in the following format:
    ///`ordinary=your.config.domain`
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub cnames: Option<Vec<String>>,

    /// specify which of the `domain` or `cnames` is
    /// the "canonical" location.
    ///
    /// this is useful for [indexing](https://developers.google.com/search/docs/crawling-indexing/consolidate-duplicate-urls)
    /// and situations where you want to display the primary
    /// URL as text on the page itself (i.e. pick one of `example.some.host`, `example.com`, and `www.example.com`).
    ///
    /// defaults to `domain` if `cnames` is empty. defaults to first `cname` in list if `cnames`
    /// are not empty.
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub canonical: Option<String>,

    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub http: Option<HttpConfig>,

    #[serde(skip)]
    #[serde(default)]
    pub internal_middlewares: Option<HashMap<String, MiddlewareConfig>>,

    /// list of email addresses that can be used to contact
    /// the application owner or administrators.
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub contacts: Option<Vec<String>>,

    /// whether contacts should be hidden (defaults to `true`)
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub hide_contacts: Option<bool>,

    /// Storage size in bytes (rounded up to nearest OS page size).
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default = "OrdinaryConfig::default_storage_size")]
    pub storage_size: Option<u64>,

    /// Specifies runtime mode for application on the host.
    ///
    /// If none is specified, defaults to Shared (or host default).
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub runtime: Option<RuntimeMode>,

    /// When set to true, `{{ domain }}/.ordinary/schema`
    /// is not addressable.
    ///
    /// Note: this can break applications which depend on
    /// flags, and `function`/`template` query descriptors.
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub hide_schema: Option<bool>,

    /// Include E2EE handler code in the client WASM.
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub client_events: Option<bool>,

    /// Port to be used for standalone "run" instances.
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub port: Option<u16>,
    /// port used for redirecting from http when
    /// standalone is running in secure mode.
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub redirect_port: Option<u16>,

    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub logging: Option<LoggingConfig>,
    /// Configures error handling.
    ///
    /// Note: If not included just the error message will be
    /// sent back as text.
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub error: Option<ErrorConfig>,
    /// Auth config for the Ordinary application.
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub auth: Option<AuthConfig>,
    /// Global constants that can be accessed from functions
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub globals: Option<BTreeMap<String, String>>,
    /// Secrets that can be used by functions.
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub secrets: Option<Vec<Secret>>,
    /// Definitions for the models that will be stored in the Ordinary database.
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub database: Option<DatabaseConfig>,
    /// IO, access and language configuration for functions that
    /// are compiled to and executed as WebAssembly modules.
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub functions: Option<Vec<FunctionConfig>>,
    /// Specifies the asset directory and per-path configuration
    /// details for assets that require preprocessing (TypeScript, SCSS,
    /// JavaScript minification, etc.)
    #[serde(skip_serializing_if = "Option::is_none")]
    #[serde(default)]
    pub assets: Option<AssetsConfig>,
}

impl OrdinaryConfig {
    /// gets ordinary.json from project path and deserializes to struct.
    pub fn get(proj_path: impl AsRef<Path>, load_refs: bool) -> anyhow::Result<OrdinaryConfig> {
        let path = proj_path.as_ref().join("ordinary.json");
        // todo: switch to async
        let mut config_json = fs::read(&path)?;

        let mut config = match simd_json::from_slice::<OrdinaryConfig>(&mut config_json) {
            Ok(config) => config,
            Err(err) => bail!("{}: {err}", path.display()),
        };

        if load_refs {
            config.load_refs(proj_path.as_ref())?;
        }

        config.load_internal();

        if let Some(database_config) = config.database.as_mut() {
            database_config.models.sort_by_key(|m| m.idx);

            for model_config in &mut database_config.models {
                model_config.fields.sort_by_key(|m| m.idx);

                for field in &mut model_config.fields {
                    field.kind.sort_sub_fields();
                }
            }
        }

        Ok(config)
    }

    pub fn write(&self, proj_path: &Path) -> anyhow::Result<()> {
        use std::io::Write;

        let ordinary_json = serde_json::to_string_pretty(self)?;

        let mut file = fs::File::create(proj_path.join("ordinary.json"))?;
        file.write_all(ordinary_json.as_bytes())?;

        Ok(())
    }

    fn load_refs(&mut self, proj_path: &Path) -> anyhow::Result<()> {
        if let Some(functions) = self.functions.as_mut() {
            Self::load_function_refs(proj_path, functions)?;
        }

        Ok(())
    }

    fn load_function_refs(
        proj_path: &Path,
        function_configs: &mut Vec<FunctionConfig>,
    ) -> anyhow::Result<()> {
        for base_function_config in function_configs {
            if let Some(reference) = &base_function_config.r#ref {
                let json_path = proj_path.join(reference);
                let mut json_bytes = fs_err::read(json_path)?;

                let mut ref_function_config: FunctionConfig =
                    simd_json::from_slice(json_bytes.as_mut_slice())?;
                ref_function_config.load_bindgen();

                ref_function_config
                    .r#ref
                    .clone_from(&base_function_config.r#ref);

                if let Some(name) = &base_function_config.name {
                    ref_function_config.name = Some(name.clone());
                }

                if let Some(timeout) = base_function_config.timeout {
                    ref_function_config.timeout = Some(timeout);
                }

                *base_function_config = ref_function_config;
            }
        }

        Ok(())
    }

    #[must_use]
    pub fn database_model_map(&self) -> HashMap<String, DatabaseModelConfig> {
        let mut model_map = HashMap::new();

        let mut database_config = self.database.clone();

        if let Some(database_config) = database_config.as_mut() {
            database_config.models.sort_by_key(|m| m.idx);

            for model_config in &mut database_config.models {
                model_config.fields.sort_by_key(|m| m.idx);

                for field in &mut model_config.fields {
                    field.kind.sort_sub_fields();
                }
            }
        }

        if let Some(database_config) = database_config {
            for model_config in &database_config.models {
                model_map.insert(model_config.name.clone(), model_config.clone());
            }
        }

        model_map
    }

    pub fn load_internal(&mut self) {
        self.canonical = Some(
            self.canonical.clone().unwrap_or(
                self.cnames
                    .clone()
                    .unwrap_or_default()
                    .first()
                    .map(ToOwned::to_owned)
                    .unwrap_or(self.domain.clone()),
            ),
        );

        self.load_internal_middlewares();
        self.load_internal_compression();
        self.load_internal_content_types();
    }

    fn load_internal_middlewares(&mut self) {
        if let Some(http_config) = &self.http
            && let Some(middlewares) = &http_config.middlewares
        {
            let mut map = HashMap::new();

            for middleware in middlewares {
                map.insert(middleware.name.clone(), middleware.clone());
            }

            self.internal_middlewares = Some(map);
        }
    }

    fn load_internal_compression(&mut self) {
        if let Some(assets) = self.assets.as_mut()
            && let Some(precompression) = &assets.precompression
        {
            assets.internal_precompression = Some(precompression.get_list());
        }

        if let Some(http_config) = self.http.as_mut()
            && let Some(http_routes) = http_config.routes.as_mut()
        {
            for http_route in http_routes {
                if let Some(http_config) = http_route.config.as_mut()
                    && let Some(http_cache) = http_config.cache.as_mut()
                    && let Some(stored_cache) = http_cache.stored.as_mut()
                    && let Some(compression) = &stored_cache.compression
                {
                    stored_cache.internal_compressions = Some(compression.get_list());
                }
            }
        }
    }

    fn load_internal_content_types(&mut self) {
        if let Some(http_config) = self.http.as_mut()
            && let Some(http_routes) = http_config.routes.as_mut()
        {
            for http_route in http_routes {
                if let Some(http_config) = http_route.config.as_mut()
                    && let Some(http_cache) = http_config.cache.as_mut()
                    && let Some(stored_cache) = http_cache.stored.as_mut()
                {
                    stored_cache.internal_content_types = Some(
                        stored_cache
                            .content_types
                            .clone()
                            .unwrap_or(smallvec!["text/html".into(), "application/json".into()]),
                    );
                }
            }
        }
    }

    #[must_use]
    pub fn get_middlewares(&self, middleware_names: &Vec<String>) -> Option<Vec<MiddlewareConfig>> {
        let mut middleware_configs = vec![];

        if let Some(middleware_map) = &self.internal_middlewares {
            for middleware in middleware_names {
                if let Some(middleware_config) = middleware_map.get(middleware) {
                    middleware_configs.push(middleware_config.clone());
                }
            }
        }

        if !middleware_configs.is_empty() {
            return Some(middleware_configs);
        }

        None
    }

    /// gets ordinary.json from project path, deserializes to struct and
    /// strips out all client-only values.
    pub fn for_send(&self) -> anyhow::Result<OrdinaryConfig> {
        let mut config = self.clone();

        config.lifecycle = None;

        if let Some(assets) = config.assets.as_mut() {
            assets.dir_path = None;
        }

        if let Some(function_configs) = config.functions.as_mut() {
            for function_config in function_configs {
                function_config.r#ref = None;

                function_config.build = None;
                function_config.bin = None;
                function_config.bindgen = None;
            }
        }

        Ok(config)
    }

    /// check that all configuration values are internally consistent
    /// and no non-existent properties or fields are used.
    #[instrument(skip_all, err, level = "debug")]
    pub fn validate(&self) -> anyhow::Result<()> {
        validate(self)
    }

    // defaults
    #[must_use]
    #[allow(clippy::unnecessary_wraps)]
    pub fn default_storage_size() -> Option<u64> {
        Some(5_000_000)
    }
    // end defaults

    /// Check that all the configuration properties are within API specified
    /// limits.
    ///
    /// Note: privileged domains are not subject to limitations checks.
    #[allow(clippy::too_many_lines)]
    pub fn check_config_against_limits(
        &self,
        limits: &OrdinaryHostLimits,
        privileged_domains: &HashSet<String>,
    ) -> anyhow::Result<()> {
        check_config_against_limits(self, limits, privileged_domains)
    }

    pub fn exec_script(
        proj_path: &Path,
        argument: &Option<String>,
        name: &str,
        when: &str,
        scripts: &Vec<Vec<String>>,
    ) -> anyhow::Result<()> {
        let span = tracing::info_span!("lifecycle", %when, %name);

        span.in_scope(|| {
            exec_script(proj_path, argument, scripts)?;
            anyhow::Ok(())
        })
    }

    pub fn check_function_name_exists(&self, name: &str) -> anyhow::Result<()> {
        if let Some(function_configs) = &self.functions {
            for function_config in function_configs {
                if function_config.name.as_deref() == Some(name) {
                    bail!("function with name {name} already exists");
                }
            }
        }

        Ok(())
    }
}

pub fn exec_script(
    proj_path: &Path,
    argument: &Option<String>,
    scripts: &Vec<Vec<String>>,
) -> anyhow::Result<()> {
    let curr_dir = env::current_dir()?;
    env::set_current_dir(proj_path)?;

    for script in scripts {
        let mut script_iter = script.iter();

        if let Some(command) = script_iter.next() {
            let mut command_str = command.clone();
            let mut command = Command::new(command);

            for arg in script_iter {
                write!(command_str, " {arg}")?;
                command.arg(arg);
            }

            tracing::info!(cmd = %command_str, "exec");

            let output = match &argument {
                Some(arg) => command.arg(arg).output()?,
                None => command.output()?,
            };

            if !output.status.success() {
                let stderr = str::from_utf8(&output.stderr)?;
                let stdout = str::from_utf8(&output.stdout)?;

                tracing::error!(%stderr, %stdout, "failed");
                bail!(stderr.to_string());
            }
        }
    }

    env::set_current_dir(curr_dir)?;

    Ok(())
}