tatara-vm 0.2.203

//! Compose a tatara-os `SystemConfig` + a tatara-init binary path into a
//! complete bootable VM manifest — kernel derivation + initrd derivation +
//! `VmSpec` ready to hand to `VfkitEmitter`.
//!
//! The goal: one call takes everything a user has typed in Lisp
//! (`(defsystem …)`, `(definit …)`, `(defvm …)`) and gives back the
//! on-disk artifacts that `vfkit --config vm.json` can boot.

use tatara_nix::derivation::{BridgeTarget, Derivation, Outputs, Source};
use tatara_nix::synth::{Artifact, MultiSynthesizer};
use tatara_os::SystemConfig;

use crate::config::{GuestKernel, GuestRootfs, Hypervisor, VmSpec};
use crate::rootfs::{InitrdFile, LinuxRootfs};
use crate::vfkit::VfkitEmitter;

/// The full set of on-disk-buildable artifacts for one VM boot.
pub struct BootManifest {
    /// Kernel derivation — bridged to `linuxPackages.kernel` by default, or
    /// whatever the caller's `SystemConfig::kernel` pointed at.
    pub kernel: Derivation,
    /// Initrd derivation — our rootfs.cpio.gz with tatara-init + init.lisp.
    pub initrd: Derivation,
    /// The VM manifest, wired to those two derivations. Use
    /// [`VfkitEmitter::with_kernel_path`] / `with_rootfs_path` to substitute
    /// the realized /nix/store paths.
    pub vm: VmSpec,
}

/// Build a `BootManifest` from a tatara-os system configuration.
///
/// - `sys` — the tatara-os `SystemConfig` (authored with `(defsystem …)`).
/// - `init_binary_path` — `/nix/store` path of the `tatara-init` binary. In
///   practice you pass the realization of `tatara.packages.${system}.init`.
/// - `vm` — VM-shape overrides (cpus, memory, network, shares). If `None`,
///   uses `VmSpec::plex_default(<hostname>)` with sensible defaults.
pub fn compose(
    sys: &SystemConfig,
    init_binary_path: impl Into<String>,
    vm: Option<VmSpec>,
) -> BootManifest {
    let init_path = init_binary_path.into();
    let hostname = sys.hostname.clone();

    // 1. Kernel — honor whatever the SystemConfig said.
    let kernel_attr = match &sys.kernel {
        tatara_os::KernelSpec::Bridge { attr_path } => attr_path.clone(),
        tatara_os::KernelSpec::Package { name } => name.clone(),
        tatara_os::KernelSpec::Custom { .. } => "linuxPackages.kernel".into(),
    };
    let kernel = Derivation {
        name: format!("kernel-{}", sanitize(&hostname)),
        version: None,
        inputs: vec![],
        source: Source::default(),
        builder: Default::default(),
        outputs: Outputs::default(),
        env: vec![],
        sandbox: Default::default(),
        bridge: Some(BridgeTarget::nixpkgs(kernel_attr)),
        nix_expr: None,
    };

    // Peek at the VM's shares so (definit :mounts …) reflects them.
    // Fall back to plex_default's empty shares when no VmSpec override given.
    let shares_for_init: Vec<crate::config::ShareSpec> = vm
        .as_ref()
        .map(|v| v.shares.clone())
        .unwrap_or_default();

    // 2. Initrd — tatara-init + init.lisp synthesized from services + shares.
    let init_config = synthesize_init_config(sys, &shares_for_init);
    let mut rootfs = LinuxRootfs::new(&init_path, init_config)
        .with_name(format!("initrd-{}", sanitize(&hostname)));
    // /etc/hostname is useful inside the guest.
    rootfs = rootfs.with_file("/etc/hostname", format!("{}\n", sys.hostname));
    // /etc files the user declared.
    for f in &sys.environment.etc_files {
        let path = if f.path.starts_with('/') {
            f.path.clone()
        } else {
            format!("/etc/{}", f.path)
        };
        rootfs.extra_files.push(InitrdFile {
            path,
            content: crate::rootfs::InitrdContent::Inline(f.content.clone()),
            mode: 0o644,
        });
    }
    // sshd — bridge openssh, bake sshd_config + authorized_keys + host key.
    if let Some(sshd) = &sys.sshd {
        rootfs = rootfs.with_sshd(sshd.clone());
    }
    // Userspace packages — each SystemConfig.packages entry is a nixpkgs
    // attr_path. Closure is baked into the initrd; bin/* lands in /bin/.
    if !sys.packages.is_empty() {
        rootfs = rootfs.with_packages(sys.packages.iter().cloned());
    }
    let initrd = rootfs.derivation();

    // 3. VmSpec — default or user-provided, but always pointed at our
    // kernel/initrd derivations via the Bridge mechanism (fleshed out by
    // `VfkitEmitter::with_*_path` at emit time when the caller has the
    // realized store paths on hand).
    let mut vm = vm.unwrap_or_else(|| VmSpec::plex_default(&hostname));
    vm.hypervisor = Hypervisor::Vfkit;
    vm.kernel = GuestKernel::Custom {
        derivation: kernel.clone(),
    };
    vm.rootfs = GuestRootfs::Image {
        derivation: initrd.clone(),
    };
    vm.initrd = Some(initrd.clone());
    if !vm
        .cmdline
        .iter()
        .any(|s| s.contains("init=/bin/tatara-init"))
    {
        vm.cmdline.push("init=/bin/tatara-init".into());
    }

    BootManifest { kernel, initrd, vm }
}

/// Derive the init.lisp content from the system's service list + the VM's
/// shares. Mirrors the shape tatara-init expects:
///   (definit :services (…) :mounts (…))
///
/// When `sys.sshd` is set, an `sshd` service is prepended automatically.
/// Each `ShareSpec` becomes a `virtiofs` mount in the init config — the
/// mount tag equals the share's guest path with non-alphanumerics replaced
/// by underscores (keeps the tag POSIX-safe while derivable from the path).
fn synthesize_init_config(sys: &SystemConfig, shares: &[crate::config::ShareSpec]) -> String {
    let mut s = format!(
        "; auto-generated by tatara-vm::boot for '{}'\n",
        sys.hostname
    );
    s.push_str(&format!("(definit\n  :name \"{}-boot\"\n", sys.hostname));

    // Prepend sshd if the SystemConfig asked for it.
    let sshd_svc = sys.sshd.as_ref().map(|sshd| {
        format!(
            "    (:name \"sshd\" :exec \"/bin/sshd -D -f /etc/ssh/sshd_config -p {port}\" :enable #t)\n",
            port = sshd.port,
        )
    });

    if sys.services.is_empty() && sshd_svc.is_none() {
        s.push_str("  :services ()\n");
    } else {
        s.push_str("  :services (\n");
        if let Some(svc) = sshd_svc {
            s.push_str(&svc);
        }
        for svc in &sys.services {
            let enable = if svc.enable { "#t" } else { "#f" };
            s.push_str(&format!(
                "    (:name \"{}\" :exec \"{}\" :enable {})\n",
                svc.name,
                svc.exec.replace('"', "\\\""),
                enable
            ));
        }
        s.push_str("  )\n");
    }

    // :mounts — one entry per declared share. mountTag is derived from
    // the guest path; tatara-vmctl uses the same derivation so the two
    // sides agree without extra coordination.
    if shares.is_empty() {
        s.push_str("  :mounts ()");
    } else {
        s.push_str("  :mounts (\n");
        for sh in shares {
            let tag = mount_tag_for_guest_path(&sh.guest);
            let opts = if sh.read_only { "ro" } else { "rw" };
            s.push_str(&format!(
                "    (:source \"{}\" :target \"{}\" :fstype \"virtiofs\" :options \"{}\")\n",
                tag, sh.guest, opts
            ));
        }
        s.push_str("  )");
    }
    s.push_str(")\n");
    s
}

/// Derive a virtiofs mount tag from a guest path. Tatara-vmctl uses the
/// same derivation at the vfkit CLI layer so tag assignments agree.
pub fn mount_tag_for_guest_path(guest: &str) -> String {
    let mut out = String::new();
    for c in guest.chars() {
        if c.is_ascii_alphanumeric() {
            out.push(c);
        } else {
            out.push('_');
        }
    }
    // Trim leading underscores so /nix/store → nix_store (not _nix_store).
    let trimmed: String = out.chars().skip_while(|c| *c == '_').collect();
    if trimmed.is_empty() {
        "share".into()
    } else {
        trimmed
    }
}

fn sanitize(s: &str) -> String {
    s.chars()
        .map(|c| {
            if c.is_ascii_alphanumeric() || c == '-' || c == '_' {
                c
            } else {
                '-'
            }
        })
        .collect()
}

// ── BootSynthesizer — one SystemConfig, full artifact tree ────────────────

/// Multi-file emitter for a full boot artifact set.
///
/// Input: one tatara-os `SystemConfig` (typically parsed from `(defsystem …)`).
/// Output: every file a user needs to reach `vfkit --config vm.json`:
///
///   - `vm.json` / `boot.sh` — the VmSpec rendered for vfkit
///   - `kernel.nix` / `initrd.nix` — standalone Nix expressions that
///     `nix build -f kernel.nix` (etc.) realize into `/nix/store` paths
///   - `init.lisp` — the supervisor config baked into the initrd
///   - `system.json` — the typed config, re-serialized for audit
///   - `README.md` — the boot instructions, keyed to the guest spec
pub struct BootSynthesizer {
    /// Path we presume tatara-init will live at inside the guest. Used only
    /// when the init binary hasn't been realized yet — real deployments
    /// substitute a `/nix/store/...-tatara-init/bin/tatara-init` path.
    pub init_binary_path: String,
    /// VmSpec overrides (cpus, memory, shares). If None, defaults by hostname.
    pub vm_override: Option<VmSpec>,
    /// Output prefix for all emitted files. Artifact paths are relative.
    pub out_prefix: String,
    /// Include busybox + applets in the initrd. Default true (matches
    /// LinuxRootfs default); set false when building on a non-Linux host
    /// without a linux-builder (e.g. naive Darwin-only dev flow).
    pub busybox: bool,
}

impl Default for BootSynthesizer {
    fn default() -> Self {
        Self {
            init_binary_path: "${pkgs.hello}/bin/hello".into(),
            vm_override: None,
            out_prefix: "boot".into(),
            busybox: true,
        }
    }
}

impl BootSynthesizer {
    pub fn new() -> Self {
        Self::default()
    }

    pub fn with_init_binary_path(mut self, p: impl Into<String>) -> Self {
        self.init_binary_path = p.into();
        self
    }

    pub fn with_out_prefix(mut self, p: impl Into<String>) -> Self {
        self.out_prefix = p.into();
        self
    }

    pub fn with_vm_override(mut self, vm: VmSpec) -> Self {
        self.vm_override = Some(vm);
        self
    }

    pub fn with_busybox(mut self, on: bool) -> Self {
        self.busybox = on;
        self
    }
}

impl MultiSynthesizer for BootSynthesizer {
    type Input = SystemConfig;

    fn generate_all(&self, cfg: &SystemConfig) -> Vec<Artifact> {
        let mut bm = compose(cfg, self.init_binary_path.clone(), self.vm_override.clone());
        // Honor the BootSynthesizer busybox flag by regenerating the initrd
        // without busybox if asked.
        if !self.busybox {
            let shares = self
                .vm_override
                .as_ref()
                .map(|v| v.shares.clone())
                .unwrap_or_default();
            let mut rootfs = crate::rootfs::LinuxRootfs::new(
                self.init_binary_path.clone(),
                synthesize_init_config(cfg, &shares),
            )
            .with_name(format!("initrd-{}", sanitize(&cfg.hostname)))
            .without_busybox();
            rootfs = rootfs.with_file("/etc/hostname", format!("{}\n", cfg.hostname));
            for f in &cfg.environment.etc_files {
                let path = if f.path.starts_with('/') {
                    f.path.clone()
                } else {
                    format!("/etc/{}", f.path)
                };
                rootfs.extra_files.push(crate::rootfs::InitrdFile {
                    path,
                    content: crate::rootfs::InitrdContent::Inline(f.content.clone()),
                    mode: 0o644,
                });
            }
            bm.initrd = rootfs.derivation();
            bm.vm.rootfs = crate::config::GuestRootfs::Image {
                derivation: bm.initrd.clone(),
            };
            bm.vm.initrd = Some(bm.initrd.clone());
        }
        let prefix = &self.out_prefix;

        // vm.json + boot.sh come from the vfkit emitter. Path placeholders
        // here — real deployments realize bm.kernel and bm.initrd and pass
        // the resulting store paths back through VfkitEmitter::with_*_path.
        let vfkit = VfkitEmitter::new();
        let mut arts = vfkit.generate_all(&bm.vm);
        // Rewrite the vfkit paths from `vm/<name>/…` to our prefix.
        for a in &mut arts {
            if let Some(suffix) = a.path.strip_prefix(&format!("vm/{}/", bm.vm.name)) {
                a.path = format!("{prefix}/{suffix}");
            }
        }

        // kernel.nix + initrd.nix — standalone buildable Nix expressions.
        // The kernel reference (e.g. `linuxPackages.kernel`) is meaningful only
        // on Linux platforms — `aarch64-linux`, `x86_64-linux`. When the boot
        // artifacts are emitted on Darwin (typical for Apple-Silicon hosts
        // using vfkit), we MUST pass `system = "${cfg.system}"` to the
        // nixpkgs import so eval picks the Linux variant of the package set.
        // Without this, `nix build -f kernel.nix` on Darwin trips the
        // `meta.platforms` assertion ("not in […linux platforms]") and
        // `launch.sh` exits before the VM ever starts.
        //
        // We only override when the bridge target has no explicit pkg_set —
        // if the user supplied one, we trust them.
        let kernel_expr = match &bm.kernel.bridge {
            Some(b) if b.pkg_set.is_none() => format!(
                "# kernel for {}\n(import <nixpkgs> {{ system = \"{}\"; }}).{}\n",
                cfg.hostname, cfg.system, b.attr_path
            ),
            Some(b) => format!(
                "# kernel for {}\n({}).{}\n",
                cfg.hostname,
                b.resolved_pkg_set(),
                b.attr_path
            ),
            None => "# (custom kernel — no bridge)\n".into(),
        };
        let initrd_expr = bm
            .initrd
            .nix_expr
            .clone()
            .unwrap_or_else(|| "# (initrd has no nix_expr — unexpected)\n".into());

        arts.push(Artifact::new(format!("{prefix}/kernel.nix"), kernel_expr));
        arts.push(Artifact::new(format!("{prefix}/initrd.nix"), initrd_expr));

        // init.lisp — extracted from the initrd expression for auditability.
        let shares_for_init = self
            .vm_override
            .as_ref()
            .map(|v| v.shares.clone())
            .unwrap_or_default();
        arts.push(Artifact::new(
            format!("{prefix}/init.lisp"),
            synthesize_init_config(cfg, &shares_for_init),
        ));

        // system.json — canonical typed view of the system.
        if let Ok(json) = serde_json::to_string_pretty(cfg) {
            arts.push(Artifact::new(format!("{prefix}/system.json"), json));
        }

        // README.md — human-friendly boot instructions.
        arts.push(Artifact::new(
            format!("{prefix}/README.md"),
            render_readme(cfg, &bm),
        ));

        arts
    }
}

fn render_readme(cfg: &SystemConfig, bm: &BootManifest) -> String {
    format!(
        "# tatara-os boot artifact — `{hostname}`\n\n\
         Generated from a `(defsystem …)` Lisp form via `tatara-vm::BootSynthesizer`.\n\n\
         ## Files\n\n\
         - `system.json`  — the typed `SystemConfig`\n\
         - `init.lisp`    — the tatara-init supervisor config (baked into the initrd)\n\
         - `kernel.nix`   — `nix build -f kernel.nix` → a Linux kernel derivation\n\
         - `initrd.nix`   — `nix build -f initrd.nix` → `{initrd_name}/initrd.cpio.gz`\n\
         - `vm.json`      — vfkit config with placeholders for the realized paths\n\
         - `boot.sh`      — helper that runs `vfkit --config vm.json`\n\n\
         ## To boot\n\n\
         ```sh\n\
         KERNEL=$(nix build -f kernel.nix --no-link --print-out-paths)/bzImage\n\
         INITRD=$(nix build -f initrd.nix --no-link --print-out-paths)/initrd.cpio.gz\n\
         # Substitute paths into vm.json (jq recommended) and run:\n\
         ./boot.sh\n\
         ```\n\n\
         ## Spec\n\n\
         - Host: `{hostname}` on `{system}`\n\
         - Init system: `{init:?}` (tatara-init is PID 1 by default)\n\
         - Services: {n_services}\n\
         - Kernel: `{kernel_name}`\n\
         - Initrd: `{initrd_name}`\n\
         - vfkit CPUs: {cpus}, memory: {mem_mib} MiB\n",
        hostname = cfg.hostname,
        system = cfg.system,
        init = cfg.init,
        n_services = cfg.services.len(),
        kernel_name = bm.kernel.name,
        initrd_name = bm.initrd.name,
        cpus = bm.vm.cpus,
        mem_mib = bm.vm.memory_mib,
    )
}

#[cfg(test)]
mod tests {
    use super::*;

    fn sys() -> SystemConfig {
        SystemConfig {
            hostname: "plex".into(),
            system: "aarch64-linux".into(),
            kernel: tatara_os::KernelSpec::Bridge {
                attr_path: "linuxPackages.kernel".into(),
            },
            bootloader: Default::default(),
            init: tatara_os::InitSystem::Tatara,
            services: vec![
                tatara_os::ServiceSpec {
                    name: "demo".into(),
                    exec: "/bin/busybox sh -c 'echo tatara'".into(),
                    enable: true,
                    extra: vec![],
                    package_refs: vec![],
                },
                tatara_os::ServiceSpec {
                    name: "disabled-one".into(),
                    exec: "/bin/disabled".into(),
                    enable: false,
                    extra: vec![],
                    package_refs: vec![],
                },
            ],
            users: vec![],
            filesystems: vec![],
            environment: Default::default(),
            packages: vec![],
            sshd: None,
        }
    }

    #[test]
    fn compose_produces_kernel_initrd_and_vm() {
        let bm = compose(&sys(), "/nix/store/xxx-tatara-init/bin/tatara-init", None);
        assert_eq!(bm.kernel.name, "kernel-plex");
        assert!(bm.kernel.bridge.is_some());
        assert_eq!(bm.initrd.name, "initrd-plex");
        assert!(bm.initrd.nix_expr.is_some());
        assert_eq!(bm.vm.name, "plex");
    }

    #[test]
    fn cmdline_gets_tatara_init_appended() {
        let mut custom = VmSpec::plex_default("plex");
        custom.cmdline = vec!["console=hvc0".into()]; // no init=
        let bm = compose(&sys(), "/nix/store/xxx-init/bin/tatara-init", Some(custom));
        assert!(bm
            .vm
            .cmdline
            .iter()
            .any(|s| s.contains("init=/bin/tatara-init")));
    }

    #[test]
    fn init_lisp_lists_enabled_services_only_as_enabled() {
        let bm = compose(&sys(), "/nix/store/xxx-init/bin/tatara-init", None);
        let expr = bm.initrd.nix_expr.unwrap();
        assert!(expr.contains("(:name \"demo\" :exec"));
        assert!(expr.contains(":enable #t"));
        assert!(expr.contains("(:name \"disabled-one\" :exec"));
        assert!(expr.contains(":enable #f"));
    }

    #[test]
    fn etc_hostname_is_included() {
        let bm = compose(&sys(), "/nix/store/xxx-init/bin/tatara-init", None);
        let expr = bm.initrd.nix_expr.unwrap();
        assert!(expr.contains("root/etc/hostname"));
        assert!(expr.contains("plex"));
    }

    // ── BootSynthesizer ────────────────────────────────────────────────

    #[test]
    fn synthesizer_emits_full_artifact_tree() {
        let s = BootSynthesizer::new().with_out_prefix("out");
        let arts = s.generate_all(&sys());
        let paths: Vec<&str> = arts.iter().map(|a| a.path.as_str()).collect();
        for expected in [
            "out/vm.json",
            "out/boot.sh",
            "out/kernel.nix",
            "out/initrd.nix",
            "out/init.lisp",
            "out/system.json",
            "out/README.md",
        ] {
            assert!(
                paths.contains(&expected),
                "missing artifact: {expected}\n got: {paths:?}"
            );
        }
    }

    #[test]
    fn synthesizer_kernel_nix_is_buildable_expression() {
        let s = BootSynthesizer::new();
        let arts = s.generate_all(&sys());
        let kernel = arts
            .iter()
            .find(|a| a.path.ends_with("kernel.nix"))
            .unwrap();
        assert!(kernel.content.contains("import <nixpkgs>"));
        assert!(kernel.content.contains(".linuxPackages.kernel"));
    }

    #[test]
    fn synthesizer_initrd_nix_is_buildable_expression() {
        let s = BootSynthesizer::new();
        let arts = s.generate_all(&sys());
        let initrd = arts
            .iter()
            .find(|a| a.path.ends_with("initrd.nix"))
            .unwrap();
        assert!(initrd.content.contains("runCommand"));
        assert!(initrd.content.contains("initrd.cpio.gz"));
        assert!(initrd.content.contains("tatara-init"));
    }

    #[test]
    fn synthesizer_readme_is_populated_from_spec() {
        let s = BootSynthesizer::new();
        let arts = s.generate_all(&sys());
        let readme = arts.iter().find(|a| a.path.ends_with("README.md")).unwrap();
        assert!(readme.content.contains("plex"));
        assert!(readme.content.contains("aarch64-linux"));
        assert!(readme.content.contains("Services: 2"));
    }

    #[test]
    fn custom_kernel_package_propagates_to_bridge() {
        let mut s = sys();
        s.kernel = tatara_os::KernelSpec::Bridge {
            attr_path: "linuxPackages_latest.kernel".into(),
        };
        let bm = compose(&s, "/nix/store/x/bin/tatara-init", None);
        assert_eq!(
            bm.kernel.bridge.unwrap().attr_path,
            "linuxPackages_latest.kernel"
        );
    }
}