jan_cli/

lib.rs

mod builtins;
mod cmdtest;
mod config;
mod cron;
mod deps;
mod inputs;
mod inspect;
mod packages;
pub mod remote;
mod runner;
mod spec_load;
mod yaml_closure;

pub use config::{load_user_config, UserConfig};
pub use runner::run_jan;
pub use spec_load::HostPlatform;

use std::collections::BTreeMap;
use std::ffi::OsString;
use std::path::{Path, PathBuf};
use std::process::Command;

use anyhow::{bail, Context, Result};
use rusqlite::Connection;
use serde::de::{self, Deserializer, Visitor};
use serde::Deserialize;
use std::fmt;

#[derive(Debug, Deserialize)]
pub struct RootSpec {
    pub metadata: Option<Metadata>,
    #[serde(default)]
    pub commands: BTreeMap<String, CommandNode>,
}

#[derive(Debug, Deserialize)]
pub struct Metadata {
    pub name: Option<String>,
    pub description: Option<String>,
}

/// Child-process environment declaration for a command node.
///
/// Two YAML shapes are accepted:
///
/// ```yaml
/// # Legacy / shorthand — all keys are public assignments
/// env:
///   FOO: bar
///
/// # Explicit sections
/// env:
///   public:
///     FOO: bar
///   private:
///     - GH_TOKEN
///   pass:
///     GH_TOKEN: github/pat
/// ```
///
/// `public` values are taken from the YAML. `private` names must already exist in
/// jan's own environment; their values are copied into the child and never stored
/// in the spec. `pass` maps an environment variable name to a `pass` store id;
/// jan runs `pass <id>` and sets only the first line of stdout as that variable
/// in the child. When any section is non-empty, the child runs with a cleared
/// environment containing only those variables plus a small essential allowlist
/// (PATH, HOME, …).
#[derive(Debug, Default, Clone, PartialEq, Eq)]
pub struct EnvSpec {
    pub public: BTreeMap<String, String>,
    pub private: Vec<String>,
    /// Env var name → `pass` store id (e.g. `GH_TOKEN` → `github/pat`).
    pub pass: BTreeMap<String, String>,
}

impl EnvSpec {
    pub fn is_empty(&self) -> bool {
        self.public.is_empty() && self.private.is_empty() && self.pass.is_empty()
    }

    /// True when the child should not inherit the full parent environment.
    pub fn restricts_child_env(&self) -> bool {
        !self.is_empty()
    }

    pub fn merge_from(&mut self, other: EnvSpec) {
        for (k, v) in other.public {
            self.public.insert(k, v);
        }
        for name in other.private {
            if !self.private.iter().any(|p| p == &name) {
                self.private.push(name);
            }
        }
        for (k, v) in other.pass {
            self.pass.insert(k, v);
        }
    }

    /// Reject overlapping private/pass names and empty keys/ids.
    pub fn validate(&self, path: &str) -> Result<()> {
        for name in &self.private {
            if name.trim().is_empty() {
                bail!("command '{path}': env.private entry must not be empty");
            }
        }
        for (env_name, pass_id) in &self.pass {
            if env_name.trim().is_empty() {
                bail!("command '{path}': env.pass key must not be empty");
            }
            if pass_id.trim().is_empty() {
                bail!("command '{path}': env.pass id for `{env_name}` must not be empty");
            }
            if self.private.iter().any(|p| p == env_name) {
                bail!(
                    "command '{path}': env var `{env_name}` cannot be both `env.private` and `env.pass`"
                );
            }
        }
        Ok(())
    }
}

impl<'de> Deserialize<'de> for EnvSpec {
    fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
    where
        D: Deserializer<'de>,
    {
        #[derive(Deserialize)]
        struct Structured {
            #[serde(default)]
            public: BTreeMap<String, String>,
            #[serde(default, deserialize_with = "deserialize_string_or_seq")]
            private: Vec<String>,
            #[serde(default)]
            pass: BTreeMap<String, String>,
        }

        #[derive(Deserialize)]
        #[serde(untagged)]
        enum EnvDe {
            Flat(BTreeMap<String, String>),
            Sections(Structured),
        }

        Ok(match EnvDe::deserialize(deserializer)? {
            EnvDe::Flat(public) => Self {
                public,
                private: Vec::new(),
                pass: BTreeMap::new(),
            },
            EnvDe::Sections(s) => Self {
                public: s.public,
                private: s.private,
                pass: s.pass,
            },
        })
    }
}

/// Whether an include link pointed at YAML (subtree graft) or a script file (exec leaf).
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum IncludeLinkKind {
    Yaml,
    Script,
}

/// Retained include identity after load (not authored directly in YAML).
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct IncludeLink {
    pub kind: IncludeLinkKind,
    /// Relative path under the jan use root (local includes).
    pub path: Option<String>,
    /// Remote URL when the include was fetched over HTTPS.
    pub url: Option<String>,
    /// Declared SHA256 when present (required for remote; optional for local).
    pub sha256: Option<String>,
}

#[derive(Debug, Deserialize, Default, Clone)]
pub struct CommandNode {
    /// If non-empty, this command and its subtree are only offered on these
    /// platforms (`linux`, `macos`, `windows`, …). `darwin` is accepted as an alias for `macos`.
    #[serde(default)]
    pub os: Vec<String>,
    #[serde(default)]
    pub about: String,
    /// Directory prepended to PATH when this script (or a descendant leaf) runs.
    pub path: Option<String>,
    /// Other script names whose `path` directories are prepended before this one runs.
    #[serde(default)]
    pub dependencies: Vec<String>,
    /// External binaries that must be on PATH (e.g. `fzf`, `jq`) before the leaf runs.
    #[serde(default)]
    pub requires: Vec<String>,
    /// Public assignments and/or private names required from the host environment.
    #[serde(default)]
    pub env: EnvSpec,
    /// Named CLI inputs (`--name value`) available as `${{ inputs.name }}` in env/argv.
    #[serde(default)]
    pub inputs: BTreeMap<String, crate::inputs::InputDef>,
    /// Optional crontab schedule(s). When set, `jan cron` runs this script's `run`
    /// leaf whenever the local time matches any expression.
    #[serde(default, deserialize_with = "deserialize_string_or_seq")]
    pub cron: Vec<String>,
    /// Package-manager dependencies (uv now; pnpm reserved).
    #[serde(default)]
    pub packages: PackagesSpec,
    /// Optional Given/When/Then shell tests (`jan test <path>`).
    #[serde(default)]
    pub tests: BTreeMap<String, CommandTest>,
    #[serde(default)]
    pub commands: BTreeMap<String, CommandNode>,
    pub exec: Option<ExecSpec>,
    /// Include link this node was loaded from, if any (filled by the loader).
    #[serde(skip)]
    pub source: Option<IncludeLink>,
}

/// One Given/When/Then shell test on a command node.
///
/// Names must follow `given_…_when_…_then_…` (spaces or hyphens are fine).
/// `when` is extra argv for the command this test is declared on (shell-expanded);
/// omit it to invoke that command with no extra args. stdout/stderr/status of that
/// invocation are captured as `JAN_STATUS` / `JAN_STDOUT` / `JAN_STDERR` for `then`.
#[derive(Debug, Clone, Default, PartialEq, Eq, Deserialize)]
pub struct CommandTest {
    /// Setup (files, fixtures). Optional.
    #[serde(default)]
    pub given: String,
    /// Extra argv after the command this test is declared on. Optional.
    #[serde(default)]
    pub when: String,
    /// Assertions against the captured `when` result and any files from `given`.
    #[serde(default)]
    pub then: String,
}

impl CommandTest {
    pub fn validate(&self, path: &str, name: &str) -> Result<()> {
        if !gherkin_test_name(name) {
            bail!(
                "command '{path}': test `{name}` must follow the given_…_when_…_then_… naming pattern"
            );
        }
        if self.then.trim().is_empty() {
            bail!("command '{path}': test `{name}` needs a non-empty `then:` script");
        }
        Ok(())
    }
}

/// True when `name` is `given_…_when_…_then_…` after normalizing spaces/hyphens.
pub fn gherkin_test_name(name: &str) -> bool {
    let n: String = name
        .trim()
        .to_ascii_lowercase()
        .chars()
        .map(|c| {
            if c == '-' || c.is_whitespace() {
                '_'
            } else {
                c
            }
        })
        .collect();
    let n = n
        .split('_')
        .filter(|s| !s.is_empty())
        .collect::<Vec<_>>()
        .join("_");
    let Some(rest) = n.strip_prefix("given_") else {
        return false;
    };
    let Some((given_body, after_when)) = rest.split_once("_when_") else {
        return false;
    };
    let Some((when_body, then_body)) = after_when.split_once("_then_") else {
        return false;
    };
    !given_body.is_empty() && !when_body.is_empty() && !then_body.is_empty()
}

/// Package-manager deps for a command node (`packages:` in YAML).
///
/// Distinct from `dependencies:`, which names other jan scripts whose `path`
/// directories are prepended to PATH.
#[derive(Debug, Clone, Default, PartialEq, Eq, Deserialize)]
pub struct PackagesSpec {
    #[serde(default)]
    pub uv: Option<UvPackages>,
    #[serde(default)]
    pub pnpm: Option<PnpmPackages>,
    #[serde(default)]
    pub gradle: Option<GradlePackages>,
}

impl PackagesSpec {
    pub fn is_empty(&self) -> bool {
        self.uv.is_none() && self.pnpm.is_none() && self.gradle.is_none()
    }

    /// Deeper node wins per manager (no list merge).
    pub fn merge_from(&mut self, other: PackagesSpec) {
        if other.uv.is_some() {
            self.uv = other.uv;
        }
        if other.pnpm.is_some() {
            self.pnpm = other.pnpm;
        }
        if other.gradle.is_some() {
            self.gradle = other.gradle;
        }
    }

    pub fn validate(&self, path: &str) -> Result<()> {
        if let Some(uv) = &self.uv {
            uv.validate(path)?;
        }
        if let Some(pnpm) = &self.pnpm {
            pnpm.validate(path)?;
        }
        if let Some(gradle) = &self.gradle {
            gradle.validate(path)?;
        }
        Ok(())
    }
}

/// uv dependency declaration: inline list, project dir, or requirements file,
/// with an optional minimum Python version (`python:`).
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct UvPackages {
    pub deps: UvDeps,
    /// Minimum Python version, e.g. `3.11` or `>=3.11`.
    pub python: Option<String>,
}

#[derive(Debug, Clone, PartialEq, Eq)]
pub enum UvDeps {
    List(Vec<String>),
    Project(String),
    Requirements(String),
}

impl UvPackages {
    pub fn list(pkgs: Vec<String>) -> Self {
        Self {
            deps: UvDeps::List(pkgs),
            python: None,
        }
    }

    pub fn validate(&self, path: &str) -> Result<()> {
        if let Some(py) = &self.python {
            packages::parse_min_version_constraint(py)
                .map_err(|e| anyhow::anyhow!("command '{path}': packages.uv.python: {e}"))?;
        }
        match &self.deps {
            UvDeps::List(pkgs) => {
                if pkgs.is_empty() {
                    bail!("command '{path}': packages.uv list must not be empty");
                }
                for p in pkgs {
                    if p.trim().is_empty() {
                        bail!("command '{path}': packages.uv entry must not be empty");
                    }
                    packages::check_pinned_requirement(p)
                        .map_err(|e| anyhow::anyhow!("command '{path}': {e}"))?;
                }
            }
            UvDeps::Project(p) | UvDeps::Requirements(p) => {
                if p.trim().is_empty() {
                    bail!("command '{path}': packages.uv path must not be empty");
                }
            }
        }
        Ok(())
    }
}

impl<'de> Deserialize<'de> for UvPackages {
    fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
    where
        D: Deserializer<'de>,
    {
        #[derive(Deserialize)]
        #[serde(deny_unknown_fields)]
        struct MapForm {
            #[serde(default)]
            project: Option<String>,
            #[serde(default)]
            requirements: Option<String>,
            #[serde(default, alias = "deps")]
            packages: Option<Vec<String>>,
            #[serde(default, deserialize_with = "deserialize_opt_stringish")]
            python: Option<String>,
        }

        #[derive(Deserialize)]
        #[serde(untagged)]
        enum Helper {
            List(Vec<String>),
            Map(MapForm),
        }

        match Helper::deserialize(deserializer)? {
            Helper::List(pkgs) => {
                let pkgs: Vec<String> = pkgs
                    .into_iter()
                    .map(|s| s.trim().to_string())
                    .filter(|s| !s.is_empty())
                    .collect();
                Ok(UvPackages {
                    deps: UvDeps::List(pkgs),
                    python: None,
                })
            }
            Helper::Map(m) => {
                let project = m
                    .project
                    .map(|s| s.trim().to_string())
                    .filter(|s| !s.is_empty());
                let requirements = m
                    .requirements
                    .map(|s| s.trim().to_string())
                    .filter(|s| !s.is_empty());
                let packages = m.packages.map(|pkgs| {
                    pkgs.into_iter()
                        .map(|s| s.trim().to_string())
                        .filter(|s| !s.is_empty())
                        .collect::<Vec<_>>()
                });
                let python = m
                    .python
                    .map(|s| s.trim().to_string())
                    .filter(|s| !s.is_empty());
                let deps = match (project, requirements, packages) {
                    (Some(p), None, None) => UvDeps::Project(p),
                    (None, Some(r), None) => UvDeps::Requirements(r),
                    (None, None, Some(pkgs)) => UvDeps::List(pkgs),
                    _ => {
                        return Err(de::Error::custom(
                            "packages.uv map must set exactly one of `packages`, `project`, or `requirements`",
                        ));
                    }
                };
                Ok(UvPackages { deps, python })
            }
        }
    }
}

/// pnpm dependency declaration: inline list or a project dir with a lockfile,
/// with an optional minimum Node version (`node:`).
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct PnpmPackages {
    pub deps: PnpmDeps,
    /// Minimum Node.js version, e.g. `18` or `>=18.0.0`.
    pub node: Option<String>,
}

#[derive(Debug, Clone, PartialEq, Eq)]
pub enum PnpmDeps {
    List(Vec<String>),
    Project(String),
}

impl PnpmPackages {
    pub fn list(pkgs: Vec<String>) -> Self {
        Self {
            deps: PnpmDeps::List(pkgs),
            node: None,
        }
    }

    pub fn validate(&self, path: &str) -> Result<()> {
        if let Some(node) = &self.node {
            packages::parse_min_version_constraint(node)
                .map_err(|e| anyhow::anyhow!("command '{path}': packages.pnpm.node: {e}"))?;
        }
        match &self.deps {
            PnpmDeps::List(pkgs) => {
                if pkgs.is_empty() {
                    bail!("command '{path}': packages.pnpm list must not be empty");
                }
                for p in pkgs {
                    if p.trim().is_empty() {
                        bail!("command '{path}': packages.pnpm entry must not be empty");
                    }
                    packages::check_pinned_npm_spec(p)
                        .map_err(|e| anyhow::anyhow!("command '{path}': {e}"))?;
                }
            }
            PnpmDeps::Project(p) => {
                if p.trim().is_empty() {
                    bail!("command '{path}': packages.pnpm path must not be empty");
                }
            }
        }
        Ok(())
    }
}

impl<'de> Deserialize<'de> for PnpmPackages {
    fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
    where
        D: Deserializer<'de>,
    {
        #[derive(Deserialize)]
        #[serde(deny_unknown_fields)]
        struct MapForm {
            #[serde(default)]
            project: Option<String>,
            #[serde(default, alias = "deps")]
            packages: Option<Vec<String>>,
            #[serde(default, deserialize_with = "deserialize_opt_stringish")]
            node: Option<String>,
        }

        #[derive(Deserialize)]
        #[serde(untagged)]
        enum Helper {
            List(Vec<String>),
            Map(MapForm),
        }

        match Helper::deserialize(deserializer)? {
            Helper::List(pkgs) => {
                let pkgs: Vec<String> = pkgs
                    .into_iter()
                    .map(|s| s.trim().to_string())
                    .filter(|s| !s.is_empty())
                    .collect();
                Ok(PnpmPackages {
                    deps: PnpmDeps::List(pkgs),
                    node: None,
                })
            }
            Helper::Map(m) => {
                let project = m
                    .project
                    .map(|s| s.trim().to_string())
                    .filter(|s| !s.is_empty());
                let packages = m.packages.map(|pkgs| {
                    pkgs.into_iter()
                        .map(|s| s.trim().to_string())
                        .filter(|s| !s.is_empty())
                        .collect::<Vec<_>>()
                });
                let node = m
                    .node
                    .map(|s| s.trim().to_string())
                    .filter(|s| !s.is_empty());
                let deps = match (project, packages) {
                    (Some(p), None) => PnpmDeps::Project(p),
                    (None, Some(pkgs)) => PnpmDeps::List(pkgs),
                    _ => {
                        return Err(de::Error::custom(
                            "packages.pnpm map must set exactly one of `packages` or `project`",
                        ));
                    }
                };
                Ok(PnpmPackages { deps, node })
            }
        }
    }
}

/// Gradle dependency declaration: pinned Maven coordinates or a locked project,
/// with an optional minimum JDK (`java:` / `jdk:`).
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct GradlePackages {
    pub deps: GradleDeps,
    /// Minimum JDK/Java version, e.g. `21` or `>=21`.
    pub java: Option<String>,
}

#[derive(Debug, Clone, PartialEq, Eq)]
pub enum GradleDeps {
    List(Vec<String>),
    Project(String),
}

impl GradlePackages {
    pub fn list(pkgs: Vec<String>) -> Self {
        Self {
            deps: GradleDeps::List(pkgs),
            java: None,
        }
    }

    pub fn validate(&self, path: &str) -> Result<()> {
        if let Some(java) = &self.java {
            packages::parse_min_version_constraint(java)
                .map_err(|e| anyhow::anyhow!("command '{path}': packages.gradle.java: {e}"))?;
        }
        match &self.deps {
            GradleDeps::List(pkgs) => {
                if pkgs.is_empty() {
                    bail!("command '{path}': packages.gradle list must not be empty");
                }
                for p in pkgs {
                    if p.trim().is_empty() {
                        bail!("command '{path}': packages.gradle entry must not be empty");
                    }
                    packages::check_pinned_maven_coord(p)
                        .map_err(|e| anyhow::anyhow!("command '{path}': {e}"))?;
                }
            }
            GradleDeps::Project(p) => {
                if p.trim().is_empty() {
                    bail!("command '{path}': packages.gradle path must not be empty");
                }
            }
        }
        Ok(())
    }
}

impl<'de> Deserialize<'de> for GradlePackages {
    fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
    where
        D: Deserializer<'de>,
    {
        #[derive(Deserialize)]
        #[serde(deny_unknown_fields)]
        struct MapForm {
            #[serde(default)]
            project: Option<String>,
            #[serde(default, alias = "deps")]
            packages: Option<Vec<String>>,
            #[serde(default, alias = "jdk", deserialize_with = "deserialize_opt_stringish")]
            java: Option<String>,
        }

        #[derive(Deserialize)]
        #[serde(untagged)]
        enum Helper {
            List(Vec<String>),
            Map(MapForm),
        }

        match Helper::deserialize(deserializer)? {
            Helper::List(pkgs) => {
                let pkgs: Vec<String> = pkgs
                    .into_iter()
                    .map(|s| s.trim().to_string())
                    .filter(|s| !s.is_empty())
                    .collect();
                Ok(GradlePackages {
                    deps: GradleDeps::List(pkgs),
                    java: None,
                })
            }
            Helper::Map(m) => {
                let project = m
                    .project
                    .map(|s| s.trim().to_string())
                    .filter(|s| !s.is_empty());
                let packages = m.packages.map(|pkgs| {
                    pkgs.into_iter()
                        .map(|s| s.trim().to_string())
                        .filter(|s| !s.is_empty())
                        .collect::<Vec<_>>()
                });
                let java = m
                    .java
                    .map(|s| s.trim().to_string())
                    .filter(|s| !s.is_empty());
                let deps = match (project, packages) {
                    (Some(p), None) => GradleDeps::Project(p),
                    (None, Some(pkgs)) => GradleDeps::List(pkgs),
                    _ => {
                        return Err(de::Error::custom(
                            "packages.gradle map must set exactly one of `packages` or `project`",
                        ));
                    }
                };
                Ok(GradlePackages { deps, java })
            }
        }
    }
}

pub(crate) fn deserialize_opt_stringish<'de, D>(deserializer: D) -> Result<Option<String>, D::Error>
where
    D: Deserializer<'de>,
{
    struct Stringish;

    impl<'de> Visitor<'de> for Stringish {
        type Value = Option<String>;

        fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result {
            formatter.write_str("a string or number version constraint, or null")
        }

        fn visit_none<E>(self) -> Result<Self::Value, E>
        where
            E: de::Error,
        {
            Ok(None)
        }

        fn visit_unit<E>(self) -> Result<Self::Value, E>
        where
            E: de::Error,
        {
            Ok(None)
        }

        fn visit_str<E>(self, value: &str) -> Result<Self::Value, E>
        where
            E: de::Error,
        {
            let t = value.trim();
            if t.is_empty() {
                Ok(None)
            } else {
                Ok(Some(t.to_string()))
            }
        }

        fn visit_string<E>(self, value: String) -> Result<Self::Value, E>
        where
            E: de::Error,
        {
            self.visit_str(&value)
        }

        fn visit_i64<E>(self, value: i64) -> Result<Self::Value, E>
        where
            E: de::Error,
        {
            Ok(Some(value.to_string()))
        }

        fn visit_u64<E>(self, value: u64) -> Result<Self::Value, E>
        where
            E: de::Error,
        {
            Ok(Some(value.to_string()))
        }

        fn visit_f64<E>(self, value: f64) -> Result<Self::Value, E>
        where
            E: de::Error,
        {
            // YAML may parse 3.11 as a float — keep a readable form.
            let s = if (value.fract()).abs() < f64::EPSILON {
                format!("{}", value as i64)
            } else {
                // Trim float noise: 3.110000 -> 3.11
                let s = format!("{value}");
                s.trim_end_matches('0').trim_end_matches('.').to_string()
            };
            Ok(Some(s))
        }
    }

    deserializer.deserialize_any(Stringish)
}

pub(crate) fn deserialize_string_or_seq<'de, D>(deserializer: D) -> Result<Vec<String>, D::Error>
where
    D: Deserializer<'de>,
{
    struct StringOrSeq;

    impl<'de> Visitor<'de> for StringOrSeq {
        type Value = Vec<String>;

        fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result {
            formatter.write_str("a string or a sequence of strings")
        }

        fn visit_str<E>(self, value: &str) -> Result<Self::Value, E>
        where
            E: de::Error,
        {
            if value.trim().is_empty() {
                Ok(Vec::new())
            } else {
                Ok(vec![value.to_string()])
            }
        }

        fn visit_string<E>(self, value: String) -> Result<Self::Value, E>
        where
            E: de::Error,
        {
            self.visit_str(&value)
        }

        fn visit_seq<A>(self, mut seq: A) -> Result<Self::Value, A::Error>
        where
            A: de::SeqAccess<'de>,
        {
            let mut out = Vec::new();
            while let Some(s) = seq.next_element::<String>()? {
                if !s.trim().is_empty() {
                    out.push(s);
                }
            }
            Ok(out)
        }

        fn visit_none<E>(self) -> Result<Self::Value, E>
        where
            E: de::Error,
        {
            Ok(Vec::new())
        }

        fn visit_unit<E>(self) -> Result<Self::Value, E>
        where
            E: de::Error,
        {
            Ok(Vec::new())
        }
    }

    deserializer.deserialize_any(StringOrSeq)
}

/// Local include under the preferred jan directory (YAML subtree or script file).
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct LocalInclude {
    pub path: String,
    /// Optional integrity pin; verified when present.
    pub sha256: Option<String>,
    /// Interpreter prefix for script includes only (e.g. `["bash"]`).
    pub argv: Vec<String>,
    /// Passthrough trailing CLI args for script includes only.
    pub passthrough: bool,
}

impl LocalInclude {
    pub fn from_path(path: impl Into<String>) -> Self {
        Self {
            path: path.into(),
            sha256: None,
            argv: Vec::new(),
            passthrough: false,
        }
    }

    pub fn is_yaml(&self) -> bool {
        let lower = self.path.to_ascii_lowercase();
        lower.ends_with(".yaml") || lower.ends_with(".yml")
    }
}

/// Local path or remote HTTPS include target.
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum IncludeRef {
    /// Relative path under the preferred jan directory (optional sha256 / script opts).
    Local(LocalInclude),
    /// Remote YAML fetched with SHA256 verification.
    Remote(RemoteInclude),
}

#[derive(Debug, Clone, Deserialize, PartialEq, Eq)]
pub struct RemoteInclude {
    pub url: String,
    pub sha256: String,
    #[serde(default)]
    pub ttl: Option<u64>,
}

impl IncludeRef {
    pub fn is_remote(&self) -> bool {
        matches!(self, Self::Remote(_))
    }

    pub fn local_path(&self) -> Option<&str> {
        match self {
            Self::Local(l) => Some(l.path.as_str()),
            Self::Remote(_) => None,
        }
    }

    pub fn cycle_token(&self) -> String {
        match self {
            Self::Local(l) => match &l.sha256 {
                Some(h) => format!("{}#{}", l.path, h.to_ascii_lowercase()),
                None => l.path.clone(),
            },
            Self::Remote(r) => format!("{}#{}", r.url, r.sha256.to_ascii_lowercase()),
        }
    }
}

impl<'de> Deserialize<'de> for IncludeRef {
    fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
    where
        D: Deserializer<'de>,
    {
        #[derive(Deserialize)]
        #[serde(deny_unknown_fields)]
        struct LocalMap {
            path: String,
            #[serde(default)]
            sha256: Option<String>,
            #[serde(default)]
            argv: Vec<String>,
            #[serde(default)]
            passthrough: bool,
        }

        #[derive(Deserialize)]
        #[serde(untagged)]
        enum Helper {
            Path(String),
            Local(LocalMap),
            Remote(RemoteInclude),
        }

        match Helper::deserialize(deserializer)? {
            Helper::Path(path) => {
                let path = path.trim();
                if path.is_empty() {
                    return Err(de::Error::custom("include path must not be empty"));
                }
                Ok(IncludeRef::Local(LocalInclude::from_path(path)))
            }
            Helper::Local(m) => {
                let path = m.path.trim();
                if path.is_empty() {
                    return Err(de::Error::custom("include.path must not be empty"));
                }
                let sha256 = m
                    .sha256
                    .map(|s| s.trim().to_string())
                    .filter(|s| !s.is_empty());
                Ok(IncludeRef::Local(LocalInclude {
                    path: path.to_string(),
                    sha256,
                    argv: m.argv,
                    passthrough: m.passthrough,
                }))
            }
            Helper::Remote(r) => {
                if r.url.trim().is_empty() {
                    return Err(de::Error::custom("include.url must not be empty"));
                }
                if r.sha256.trim().is_empty() {
                    return Err(de::Error::custom(
                        "include.sha256 is required with include.url",
                    ));
                }
                Ok(IncludeRef::Remote(r))
            }
        }
    }
}

#[derive(Debug, Deserialize, Clone, Default)]
pub struct ExecSpec {
    /// Program argv. For remote `url` / local `file` leaves this is an optional
    /// interpreter prefix (e.g. `["python3"]`); the script path is appended automatically.
    /// For `kotlin:` leaves this is the argument list passed to `main`.
    #[serde(default)]
    pub argv: Vec<String>,
    /// Append extra CLI arguments after those from `argv` / the script path.
    #[serde(default)]
    pub passthrough: bool,
    /// HTTPS URL of a remote script to download, verify, and run.
    #[serde(default)]
    pub url: Option<String>,
    /// Local script path relative to the jan use root (optional `sha256` pin).
    #[serde(default)]
    pub file: Option<String>,
    /// Kotlin source relative to the jan use root (`.kt` / `.kts`), or an inline
    /// program (multiline YAML string). Paths are a single line ending in
    /// `.kt`/`.kts`; anything else is treated as inlined source and compile-once
    /// as `.kt`. `argv` is passed to `main`.
    #[serde(default)]
    pub kotlin: Option<String>,
    /// Python source relative to the jan use root (`.py`), or an inline program.
    /// Runs with the `packages.uv` venv interpreter when present. `argv` is
    /// forwarded after the script / `-c` body (`sys.argv`).
    #[serde(default)]
    pub python: Option<String>,
    /// JavaScript source relative to the jan use root (`.js` / `.mjs` / `.cjs`),
    /// or an inline program. Runs with `node` and `packages.pnpm` `NODE_PATH`
    /// when present. `argv` is forwarded on `process.argv`.
    #[serde(default)]
    pub node: Option<String>,
    /// Bash source relative to the jan use root (`.sh` / `.bash`), or an inline
    /// program (`bash -lc`). `argv` is `$1`… (`$0` is the command name).
    #[serde(default)]
    pub bash: Option<String>,
    /// POSIX `sh` source (`.sh`) or inline (`sh -c`). `argv` is `$1`….
    #[serde(default)]
    pub sh: Option<String>,
    /// Zsh source (`.zsh` / `.sh`) or inline (`zsh -c`). `argv` is `$1`….
    #[serde(default)]
    pub zsh: Option<String>,
    /// Literal text to print with no subprocess (`text:` or alias `cat:`).
    #[serde(default, alias = "cat")]
    pub text: Option<String>,
    /// SHA256 of the script: required with `url`, optional with `file`.
    #[serde(default)]
    pub sha256: Option<String>,
    /// Optional TTL override (seconds) for the remote script cache.
    #[serde(default)]
    pub ttl: Option<u64>,
}

impl ExecSpec {
    pub fn is_remote(&self) -> bool {
        self.url
            .as_deref()
            .map(|u| !u.trim().is_empty())
            .unwrap_or(false)
    }

    pub fn is_local_file(&self) -> bool {
        self.file
            .as_deref()
            .map(|u| !u.trim().is_empty())
            .unwrap_or(false)
    }

    pub fn is_kotlin(&self) -> bool {
        self.kotlin
            .as_deref()
            .map(|u| !u.trim().is_empty())
            .unwrap_or(false)
    }

    pub fn is_python(&self) -> bool {
        self.python
            .as_deref()
            .map(|u| !u.trim().is_empty())
            .unwrap_or(false)
    }

    pub fn is_node(&self) -> bool {
        self.node
            .as_deref()
            .map(|u| !u.trim().is_empty())
            .unwrap_or(false)
    }

    pub fn is_bash(&self) -> bool {
        self.bash
            .as_deref()
            .map(|u| !u.trim().is_empty())
            .unwrap_or(false)
    }

    pub fn is_sh(&self) -> bool {
        self.sh
            .as_deref()
            .map(|u| !u.trim().is_empty())
            .unwrap_or(false)
    }

    pub fn is_zsh(&self) -> bool {
        self.zsh
            .as_deref()
            .map(|u| !u.trim().is_empty())
            .unwrap_or(false)
    }

    pub fn is_text(&self) -> bool {
        self.text
            .as_deref()
            .map(|u| !u.trim().is_empty())
            .unwrap_or(false)
    }

    /// Body for `exec.text` / `exec.cat`, trimmed of surrounding whitespace.
    pub fn literal_text(&self) -> Option<&str> {
        self.text
            .as_deref()
            .map(str::trim)
            .filter(|s| !s.is_empty())
    }

    /// True when a language `exec.*` field is set.
    pub fn is_language_source(&self) -> bool {
        self.is_kotlin()
            || self.is_python()
            || self.is_node()
            || self.is_bash()
            || self.is_sh()
            || self.is_zsh()
    }

    /// True when `raw` is a single-line `.kt` / `.kts` path (not inline source).
    pub fn kotlin_value_is_path(raw: &str) -> bool {
        Self::single_line_ext(raw, &[".kt", ".kts"])
    }

    pub fn python_value_is_path(raw: &str) -> bool {
        Self::single_line_ext(raw, &[".py"])
    }

    pub fn node_value_is_path(raw: &str) -> bool {
        Self::single_line_ext(raw, &[".js", ".mjs", ".cjs"])
    }

    pub fn bash_value_is_path(raw: &str) -> bool {
        Self::single_line_ext(raw, &[".sh", ".bash"])
    }

    pub fn sh_value_is_path(raw: &str) -> bool {
        Self::single_line_ext(raw, &[".sh"])
    }

    pub fn zsh_value_is_path(raw: &str) -> bool {
        Self::single_line_ext(raw, &[".zsh", ".sh"])
    }

    fn single_line_ext(raw: &str, exts: &[&str]) -> bool {
        let t = raw.trim();
        if t.is_empty() || t.lines().nth(1).is_some() {
            return false;
        }
        let lower = t.to_ascii_lowercase();
        exts.iter().any(|e| lower.ends_with(e))
    }

    /// True when `exec.kotlin` names a tree-relative `.kt` / `.kts` file.
    pub fn kotlin_is_path(&self) -> bool {
        self.kotlin
            .as_deref()
            .map(Self::kotlin_value_is_path)
            .unwrap_or(false)
    }

    pub fn validate(&self, path: &str) -> Result<()> {
        let url = self.url.as_deref().map(str::trim).filter(|s| !s.is_empty());
        let file = self
            .file
            .as_deref()
            .map(str::trim)
            .filter(|s| !s.is_empty());
        let kotlin = self
            .kotlin
            .as_deref()
            .map(str::trim)
            .filter(|s| !s.is_empty());
        let python = self
            .python
            .as_deref()
            .map(str::trim)
            .filter(|s| !s.is_empty());
        let node = self
            .node
            .as_deref()
            .map(str::trim)
            .filter(|s| !s.is_empty());
        let bash = self
            .bash
            .as_deref()
            .map(str::trim)
            .filter(|s| !s.is_empty());
        let sh = self.sh.as_deref().map(str::trim).filter(|s| !s.is_empty());
        let zsh = self.zsh.as_deref().map(str::trim).filter(|s| !s.is_empty());
        let text = self
            .text
            .as_deref()
            .map(str::trim)
            .filter(|s| !s.is_empty());
        let hash = self
            .sha256
            .as_deref()
            .map(str::trim)
            .filter(|s| !s.is_empty());
        let exclusive = [
            ("url", url),
            ("file", file),
            ("kotlin", kotlin),
            ("python", python),
            ("node", node),
            ("bash", bash),
            ("sh", sh),
            ("zsh", zsh),
            ("text", text),
        ];
        let set: Vec<(&str, &str)> = exclusive
            .iter()
            .copied()
            .filter_map(|(n, v)| v.map(|s| (n, s)))
            .collect();
        if set.len() > 1 {
            bail!(
                "command '{path}': exec cannot combine `url`, `file`, `kotlin`, `python`, `node`, `bash`, `sh`, `zsh`, and `text`"
            );
        }
        const LANG: &[&str] = &["kotlin", "python", "node", "bash", "sh", "zsh"];
        if hash.is_some() && set.iter().any(|(n, _)| LANG.contains(n) || *n == "text") {
            bail!(
                "command '{path}': exec.sha256 is not supported with exec.kotlin / exec.python / exec.node / exec.bash / exec.sh / exec.zsh / exec.text"
            );
        }
        match set.first().copied() {
            Some(("url", _)) if hash.is_some() => Ok(()),
            Some(("url", _)) => {
                bail!("command '{path}': exec.sha256 is required with exec.url")
            }
            Some(("file", _)) => Ok(()),
            Some(("text", _)) => Ok(()),
            Some(("kotlin", k)) => {
                if Self::kotlin_value_is_path(k) {
                    return Ok(());
                }
                if !k.contains("fun ") && !k.contains("fun\t") {
                    bail!(
                        "command '{path}': exec.kotlin inline source must contain a `fun` \
                         (or set a single-line `.kt` / `.kts` path)"
                    );
                }
                Ok(())
            }
            Some((label, src)) if LANG.contains(&label) => {
                let is_path = match label {
                    "python" => Self::python_value_is_path(src),
                    "node" => Self::node_value_is_path(src),
                    "bash" => Self::bash_value_is_path(src),
                    "sh" => Self::sh_value_is_path(src),
                    "zsh" => Self::zsh_value_is_path(src),
                    _ => false,
                };
                if is_path {
                    return Ok(());
                }
                if src.len() < 2 {
                    bail!("command '{path}': exec.{label} inline source is empty");
                }
                Ok(())
            }
            None if hash.is_some() => {
                bail!("command '{path}': exec.sha256 requires exec.url or exec.file")
            }
            None => {
                if self.argv.is_empty() {
                    bail!(
                        "command '{path}': exec.argv must not be empty (or set exec.url / exec.file / exec.kotlin / exec.python / exec.node / exec.bash / exec.sh / exec.zsh / exec.text)"
                    );
                }
                Ok(())
            }
            _ => unreachable!("modes > 1 checked above"),
        }
    }
}

impl CommandNode {
    pub fn is_leaf_exec(&self) -> bool {
        self.exec.is_some()
    }

    pub fn validate(&self, path: &str) -> Result<()> {
        if self.exec.is_some() && !self.commands.is_empty() {
            bail!("command '{path}' cannot define both `exec` and nested `commands`");
        }
        if let Some(ref e) = self.exec {
            e.validate(path)?;
        }
        self.env.validate(path)?;
        self.packages.validate(path)?;
        for (name, t) in &self.tests {
            t.validate(path, name)?;
        }
        for (name, def) in &self.inputs {
            def.validate(name)
                .map_err(|e| anyhow::anyhow!("command '{path}': {e}"))?;
        }
        for (name, child) in &self.commands {
            let p = if path.is_empty() {
                name.clone()
            } else {
                format!("{path} {name}")
            };
            child.validate(&p)?;
        }
        Ok(())
    }
}

/// Deep-merge `overlay.commands` into `base`, letting included YAML fragments
/// add or replace leaves and extend nested groups.
pub fn merge_specs_into(base: &mut RootSpec, overlay: RootSpec) -> Result<()> {
    for (name, node) in overlay.commands {
        match base.commands.get_mut(&name) {
            Some(existing) => merge_command_node(existing, node)?,
            None => {
                base.commands.insert(name, node);
            }
        }
    }
    Ok(())
}

fn merge_command_node(dst: &mut CommandNode, src: CommandNode) -> Result<()> {
    if src.exec.is_some() && !src.commands.is_empty() {
        bail!("merge overlay: command cannot define both `exec` and nested `commands`");
    }
    if !src.os.is_empty() {
        dst.os = src.os;
    }
    if !src.about.trim().is_empty() {
        dst.about = src.about;
    }
    if src.path.is_some() {
        dst.path = src.path;
    }
    if !src.dependencies.is_empty() {
        dst.dependencies = src.dependencies;
    }
    if !src.requires.is_empty() {
        dst.requires = src.requires;
    }
    if !src.cron.is_empty() {
        dst.cron = src.cron;
    }
    if !src.env.is_empty() {
        dst.env.merge_from(src.env);
    }
    for (k, v) in src.inputs {
        dst.inputs.insert(k, v);
    }
    for (k, v) in src.tests {
        dst.tests.insert(k, v);
    }
    if let Some(exec) = src.exec {
        dst.exec = Some(exec);
        dst.commands.clear();
        return Ok(());
    }
    if !src.commands.is_empty() {
        dst.exec = None;
        for (k, child) in src.commands {
            match dst.commands.get_mut(&k) {
                Some(existing) => merge_command_node(existing, child)?,
                None => {
                    dst.commands.insert(k, child);
                }
            }
        }
    }
    Ok(())
}

/// Validate every command in the tree (after merges or programmatic edits).
pub fn validate_spec(spec: &RootSpec) -> Result<()> {
    for (name, node) in &spec.commands {
        node.validate(name)?;
    }
    Ok(())
}

/// Parse YAML from memory. Use `include_base` when the document uses `include:` (root or subtree).
pub fn load_spec_from_str(raw: &str, include_base: Option<&Path>) -> Result<RootSpec> {
    spec_load::load_spec_from_str(raw, include_base, HostPlatform::detect())
}

pub fn load_spec(path: &Path) -> Result<RootSpec> {
    spec_load::load_spec_from_path(path, HostPlatform::detect())
}

pub fn resolve_git_branch(cwd: &Path, override_branch: Option<&str>) -> String {
    if let Some(b) = override_branch {
        if !b.is_empty() {
            return b.to_string();
        }
    }
    if let Ok(v) = std::env::var("JAN_BRANCH") {
        if !v.is_empty() {
            return v;
        }
    }
    let output = Command::new("git")
        .args(["rev-parse", "--abbrev-ref", "HEAD"])
        .current_dir(cwd)
        .output();
    match output {
        Ok(o) if o.status.success() => String::from_utf8_lossy(&o.stdout).trim().to_string(),
        _ => "(no-git)".to_string(),
    }
}

fn first_line(s: &str) -> String {
    s.lines().next().unwrap_or("").trim().to_string()
}

/// Conventional `commands.help` leaf: documentation inlined into `--help`.
fn is_help_leaf(name: &str, child: &CommandNode) -> bool {
    name == "help" && child.exec.is_some() && child.commands.is_empty()
}

fn node_at_chain<'a>(spec: &'a RootSpec, chain: &[String]) -> Option<&'a CommandNode> {
    let mut map = &spec.commands;
    let mut node = None;
    for seg in chain {
        let next = map.get(seg)?;
        node = Some(next);
        map = &next.commands;
    }
    node
}

/// Documentation body for `--help`: `exec.text` on this node, else `commands.help`,
/// else the parent script's `help` when this node is the `run` leaf.
fn command_help_text(
    spec: &RootSpec,
    chain: &[String],
    node: Option<&CommandNode>,
) -> Option<String> {
    let n = node?;
    if let Some(t) = n.exec.as_ref().and_then(ExecSpec::literal_text) {
        return Some(t.to_string());
    }
    if let Some(t) = n
        .commands
        .get("help")
        .and_then(|h| h.exec.as_ref())
        .and_then(ExecSpec::literal_text)
    {
        return Some(t.to_string());
    }
    if chain.last().map(String::as_str) == Some("run") && chain.len() >= 2 {
        let parent = node_at_chain(spec, &chain[..chain.len() - 1])?;
        return parent
            .commands
            .get("help")
            .and_then(|h| h.exec.as_ref())
            .and_then(ExecSpec::literal_text)
            .map(str::to_string);
    }
    None
}

pub fn format_help(spec: &RootSpec, chain: &[String], node: Option<&CommandNode>) -> String {
    let mut out = String::new();
    let bin = spec
        .metadata
        .as_ref()
        .and_then(|m| m.name.as_deref())
        .unwrap_or("jan");
    let full_cmd = if chain.is_empty() {
        bin.to_string()
    } else {
        format!("{} {}", bin, chain.join(" "))
    };

    let (about, children, exec) = match node {
        Some(n) => (n.about.as_str(), &n.commands, n.exec.as_ref()),
        None => ("", &spec.commands, None),
    };

    if chain.is_empty() {
        if let Some(meta) = &spec.metadata {
            if let Some(desc) = &meta.description {
                out.push_str(desc.trim());
                out.push_str("\n\n");
            }
        }
    }

    let help_doc = command_help_text(spec, chain, node);
    if let Some(doc) = &help_doc {
        out.push_str(doc);
        out.push_str("\n\n");
    } else if !about.is_empty() {
        out.push_str(about.trim());
        out.push_str("\n\n");
    }

    let listed: Vec<(&String, &CommandNode)> = children
        .iter()
        .filter(|(name, child)| !is_help_leaf(name, child))
        .collect();

    if exec.is_some() && children.is_empty() {
        if help_doc.is_none() {
            out.push_str("This command runs an external program (see spec `exec.argv`).\n");
        }
        append_help_inputs_and_tests(&mut out, spec, chain, node);
        return out;
    }

    if !listed.is_empty() {
        out.push_str("Subcommands:\n");
        for (name, child) in &listed {
            let line = if child.about.is_empty() {
                format!("  {name}\n")
            } else {
                format!("  {name} — {}\n", first_line(&child.about))
            };
            out.push_str(&line);
        }
        out.push('\n');
        if listed.iter().any(|(n, _)| n.as_str() != "run") {
            out.push_str(&format!(
                "Use `{} --help` for more about a subcommand.\n",
                full_cmd
            ));
        }
        append_help_inputs_and_tests(&mut out, spec, chain, node);
    } else if exec.is_none() && help_doc.is_none() {
        out.push_str("(No subcommands defined.)\n");
        append_help_inputs_and_tests(&mut out, spec, chain, node);
    } else {
        append_help_inputs_and_tests(&mut out, spec, chain, node);
    }
    if chain.is_empty() && node.is_none() {
        out.push_str(
            "\nPlace `--help` or `-h` right after the subcommand prefix you want. Built-ins: `use`, `bundle`, `alias`, `list`, `search`, `show`, `validate`, `audit`, `cron`, `test`.\n",
        );
    }
    out
}

fn append_help_inputs_and_tests(
    out: &mut String,
    spec: &RootSpec,
    chain: &[String],
    node: Option<&CommandNode>,
) {
    let defs = inputs::collect_chain_inputs(chain, spec);
    if !defs.is_empty() {
        out.push('\n');
        out.push_str(&inputs::format_inputs_help(&defs));
    }
    let n = match node {
        Some(n) => cmdtest::count_tests(n),
        None => spec.commands.values().map(cmdtest::count_tests).sum(),
    };
    if n > 0 {
        let hint = if chain.is_empty() {
            "jan test".to_string()
        } else {
            format!("jan test {}", chain.join(" "))
        };
        out.push_str(&format!("\n{n} test(s) — run with `{hint}`.\n"));
    }
}

/// Stable identity for where a YAML spec tree is rooted (directory of linked fragments + entry file).
#[derive(Debug, Clone)]
pub struct SpecRootIdentity {
    /// Canonical directory containing top-level YAML fragments.
    pub spec_dir: String,
    /// Entry YAML file name relative to `spec_dir`.
    pub root_yaml: String,
}

/// Resolve the preferred jan directory saved by `jan use`.
pub fn resolve_preferred_spec() -> Result<(PathBuf, SpecRootIdentity)> {
    let cfg = config::load_user_config().context("load user config")?;
    let Some(dir_s) = cfg
        .jan_dir
        .as_ref()
        .map(|s| s.trim())
        .filter(|s| !s.is_empty())
    else {
        bail!(
            "no preferred jan directory configured\n\
             Run `jan use <DIR>` to save a YAML command tree (see docs/PORTABLE_SCRIPTS.md)"
        );
    };
    let dir = PathBuf::from(dir_s);
    if !dir.is_dir() {
        bail!(
            "preferred jan directory does not exist: {}\n\
             Fix the path or run `jan use <DIR>` again (config: {})",
            dir.display(),
            config::config_path().display()
        );
    }
    let root = cfg
        .spec_root
        .as_deref()
        .map(str::trim)
        .filter(|s| !s.is_empty())
        .unwrap_or("scripts.spec.yaml");
    resolve_spec_dir_entry(&dir, root)
}

/// Resolve a jan directory + entry file name into an absolute spec path and identity.
pub fn resolve_spec_dir_entry(
    spec_dir: &Path,
    root_yaml: &str,
) -> Result<(PathBuf, SpecRootIdentity)> {
    let rel = Path::new(root_yaml);
    if rel.is_absolute() {
        bail!("entry YAML must be a relative file name, not an absolute path");
    }
    if rel
        .components()
        .any(|c| matches!(c, std::path::Component::ParentDir))
    {
        bail!("entry YAML must not contain `..`");
    }
    let normal_only = rel
        .components()
        .all(|c| matches!(c, std::path::Component::Normal(_)));
    let n = rel
        .components()
        .filter(|c| matches!(c, std::path::Component::Normal(_)))
        .count();
    if !normal_only || n != 1 {
        bail!("entry YAML must be a single file name inside the jan directory");
    }
    let dir = spec_dir
        .canonicalize()
        .with_context(|| format!("canonicalize jan directory {}", spec_dir.display()))?;
    if !dir.is_dir() {
        bail!("not a directory: {}", dir.display());
    }
    let spec_path = dir.join(rel);
    if !spec_path.is_file() {
        bail!(
            "spec entry not found: {} (under {})\n\
             Expected a properly formatted jan directory (e.g. scripts.spec.yaml)",
            spec_path.display(),
            dir.display()
        );
    }
    let identity = SpecRootIdentity {
        spec_dir: dir.to_string_lossy().into_owned(),
        root_yaml: rel
            .file_name()
            .expect("relative root has file_name")
            .to_string_lossy()
            .into_owned(),
    };
    Ok((spec_path, identity))
}

pub struct RunContext<'a> {
    pub cwd: &'a Path,
    pub db_path: Option<&'a Path>,
    pub branch: String,
    pub no_log: bool,
    pub spec_root: &'a SpecRootIdentity,
}

/// True when `argv` is a POSIX-shell inline (`bash`/`zsh`/`sh`/… + `-c`/`-lc` + body)
/// with no `$0` placeholder after the body yet.
///
/// For those interpreters the first word after the `-c` string becomes `$0`, not `$1`.
/// Inlined jan scripts expect normal script semantics (`$1` / `"$@"` = user args), so
/// passthrough must insert a `$0` before forwarding.
fn shell_inline_c_needs_argv0(argv: &[String]) -> bool {
    if argv.len() != 3 {
        return false;
    }
    let prog = Path::new(&argv[0])
        .file_name()
        .and_then(|s| s.to_str())
        .unwrap_or(argv[0].as_str());
    let is_shell = matches!(
        prog,
        "bash" | "zsh" | "sh" | "dash" | "ksh" | "ash" | "busybox"
    );
    is_shell && matches!(argv[1].as_str(), "-c" | "-lc")
}

fn shell_passthrough_argv0(chain: &[String]) -> String {
    chain
        .iter()
        .rev()
        .find(|s| s.as_str() != "run")
        .cloned()
        .or_else(|| chain.last().cloned())
        .unwrap_or_else(|| "jan".to_string())
}

pub fn run_matched(
    spec: &RootSpec,
    chain: &[String],
    node: &CommandNode,
    trailing: &[OsString],
    ctx: &RunContext<'_>,
) -> Result<i32> {
    let exec = match &node.exec {
        Some(e) => e,
        None => {
            let help = format_help(spec, chain, Some(node));
            print!("{help}");
            bail!("missing subcommand");
        }
    };
    exec.validate(&chain.join(" "))?;

    if exec.is_text() {
        let body = exec.literal_text().unwrap_or("");
        println!("{body}");
        return Ok(0);
    }

    let input_defs = inputs::collect_chain_inputs(chain, spec);
    let (input_vals, rest) = inputs::resolve_inputs(&input_defs, trailing, Some(ctx.cwd))?;

    let mut argv: Vec<String> = Vec::with_capacity(exec.argv.len() + 1);
    for a in &exec.argv {
        argv.push(inputs::interpolate(a, &input_vals)?);
    }

    if exec.is_remote() {
        let url = exec.url.as_deref().unwrap().trim();
        let hash = exec.sha256.as_deref().unwrap().trim();
        let mut opts = remote::FetchOpts::new();
        if let Some(ttl) = exec.ttl {
            opts = opts.with_ttl(ttl);
        }
        let cached = remote::fetch_verified(url, hash, &opts, true)?;
        argv.push(cached.to_string_lossy().into_owned());
    } else if exec.is_local_file() {
        let rel = exec.file.as_deref().unwrap().trim();
        let use_root = Path::new(&ctx.spec_root.spec_dir);
        let resolved = spec_load::resolve_under_use_root(use_root, rel)?;
        if let Some(hash) = exec
            .sha256
            .as_deref()
            .map(str::trim)
            .filter(|s| !s.is_empty())
        {
            remote::verify_file_sha256(&resolved, hash)
                .with_context(|| format!("verify exec.file `{rel}`"))?;
        }
        argv.push(resolved.to_string_lossy().into_owned());
    } else if exec.is_language_source() {
        // Args in `argv` are for the language entrypoint; the runner is built after packages.
    } else if argv.is_empty() {
        bail!("exec.argv must not be empty");
    }

    if exec.passthrough {
        let mut rest = rest;
        // `--` after the leaf is the usual jan separator; drop one leading `--` so
        // `run -- arg` and `run arg` match for both inline shells and `exec.file` /
        // script includes. A literal first arg of `--` needs `run -- --`.
        if rest.first().is_some_and(|a| a == "--") {
            rest = rest[1..].to_vec();
        }
        if shell_inline_c_needs_argv0(&argv) {
            argv.push(shell_passthrough_argv0(chain));
        }
        for a in &rest {
            argv.push(a.to_string_lossy().into_owned());
        }
    } else if !rest.is_empty() {
        let preview = rest
            .iter()
            .take(3)
            .map(|s| s.to_string_lossy().into_owned())
            .collect::<Vec<_>>()
            .join(" ");
        bail!(
            "unexpected trailing arguments: {preview}{}",
            if rest.len() > 3 { "…" } else { "" }
        );
    }

    let cmd_path = if chain.is_empty() {
        "(root)".to_string()
    } else {
        chain.join(" ")
    };

    let (_, requires, _) = deps::collect_chain_metadata(chain, spec);
    deps::check_requires(&requires)?;

    let pkgs = packages::collect_chain_packages(chain, spec);
    let pkg_envs = packages::ensure_packages(&pkgs, ctx)?;

    if exec.is_kotlin() {
        let rel = exec.kotlin.as_deref().unwrap().trim();
        let use_root = Path::new(&ctx.spec_root.spec_dir);
        let main_args = std::mem::take(&mut argv);
        argv = packages::prepare_kotlin_argv(use_root, rel, &pkg_envs, &main_args)?;
    } else if exec.is_python() {
        let src = exec.python.as_deref().unwrap().trim();
        let use_root = Path::new(&ctx.spec_root.spec_dir);
        let main_args = std::mem::take(&mut argv);
        argv = packages::prepare_python_argv(use_root, src, &main_args)?;
    } else if exec.is_node() {
        let src = exec.node.as_deref().unwrap().trim();
        let use_root = Path::new(&ctx.spec_root.spec_dir);
        let main_args = std::mem::take(&mut argv);
        argv = packages::prepare_node_argv(use_root, src, &main_args)?;
    } else if exec.is_bash() || exec.is_sh() || exec.is_zsh() {
        let (kind, src) = if exec.is_bash() {
            (
                packages::ShellKind::Bash,
                exec.bash.as_deref().unwrap().trim(),
            )
        } else if exec.is_zsh() {
            (
                packages::ShellKind::Zsh,
                exec.zsh.as_deref().unwrap().trim(),
            )
        } else {
            (packages::ShellKind::Sh, exec.sh.as_deref().unwrap().trim())
        };
        let use_root = Path::new(&ctx.spec_root.spec_dir);
        let main_args = std::mem::take(&mut argv);
        let argv0 = shell_passthrough_argv0(chain);
        argv = packages::prepare_shell_argv(kind, use_root, src, &argv0, &main_args)?;
    } else {
        packages::inject_jvm_classpath(&mut argv, &pkg_envs);
    }

    let path_dirs = deps::resolve_path_prefixes(spec, chain, ctx)?;
    let program = packages::resolve_program_with_envs(&argv[0], &pkg_envs, &path_dirs)?;
    let mut env_spec = deps::collect_chain_env(chain, spec);
    for value in env_spec.public.values_mut() {
        *value = inputs::interpolate(value, &input_vals)?;
    }
    deps::check_private_env(&env_spec.private)?;
    let mut path_override = if !path_dirs.is_empty() {
        Some(deps::prepend_path_env(&path_dirs)?)
    } else {
        None
    };
    if !pkg_envs.is_empty() {
        path_override = Some(packages::prepend_env_paths(&pkg_envs, path_override)?);
    }
    if let Some(node_path) = packages::node_path_for(&pkg_envs) {
        env_spec
            .public
            .entry("NODE_PATH".to_string())
            .or_insert(node_path);
    }
    if let Some(classpath) = packages::classpath_for(&pkg_envs) {
        env_spec
            .public
            .entry("CLASSPATH".to_string())
            .or_insert(classpath);
    }

    let mut c = Command::new(&program);
    if argv.len() > 1 {
        c.args(&argv[1..]);
    }
    c.current_dir(ctx.cwd);
    deps::apply_process_env(&mut c, &env_spec, path_override)?;

    let status = c.status().with_context(|| format!("spawn `{}`", argv[0]))?;
    let code = status.code().unwrap_or(255);

    if !ctx.no_log {
        if let Some(db) = ctx.db_path {
            log_invocation(
                db,
                &ctx.branch,
                ctx.cwd,
                &cmd_path,
                &argv,
                code,
                ctx.spec_root,
            )?;
        }
    }

    Ok(code)
}

fn ensure_invocations_spec_root_column(conn: &Connection) -> Result<()> {
    let mut stmt = conn.prepare("PRAGMA table_info(invocations)")?;
    let cols: Vec<String> = stmt
        .query_map([], |row| row.get::<_, String>(1))?
        .collect::<std::result::Result<_, _>>()?;
    if !cols.iter().any(|c| c == "spec_root_id") {
        conn.execute(
            "ALTER TABLE invocations ADD COLUMN spec_root_id INTEGER",
            [],
        )?;
    }
    Ok(())
}

fn upsert_spec_root(conn: &Connection, spec: &SpecRootIdentity) -> Result<i64> {
    let ts = unix_ts();
    conn.execute(
        r"INSERT INTO spec_roots (spec_dir, root_yaml, last_used_ts) VALUES (?1, ?2, ?3)
          ON CONFLICT(spec_dir, root_yaml) DO UPDATE SET last_used_ts = excluded.last_used_ts",
        rusqlite::params![&spec.spec_dir, &spec.root_yaml, &ts],
    )?;
    let id: i64 = conn.query_row(
        "SELECT id FROM spec_roots WHERE spec_dir = ?1 AND root_yaml = ?2",
        [&spec.spec_dir, &spec.root_yaml],
        |r| r.get(0),
    )?;
    Ok(id)
}

fn log_invocation(
    db_path: &Path,
    branch: &str,
    cwd: &Path,
    command_path: &str,
    argv: &[String],
    exit_code: i32,
    spec_root: &SpecRootIdentity,
) -> Result<()> {
    if let Some(parent) = db_path.parent() {
        std::fs::create_dir_all(parent).ok();
    }
    let conn = Connection::open(db_path).with_context(|| format!("open {}", db_path.display()))?;
    conn.execute_batch(
        r"
        CREATE TABLE IF NOT EXISTS spec_roots (
            id INTEGER PRIMARY KEY AUTOINCREMENT,
            spec_dir TEXT NOT NULL,
            root_yaml TEXT NOT NULL,
            last_used_ts TEXT NOT NULL,
            UNIQUE(spec_dir, root_yaml)
        );
        CREATE TABLE IF NOT EXISTS invocations (
            id INTEGER PRIMARY KEY AUTOINCREMENT,
            ts TEXT NOT NULL,
            git_branch TEXT NOT NULL,
            cwd TEXT NOT NULL,
            command_path TEXT NOT NULL,
            argv_json TEXT NOT NULL,
            exit_code INTEGER NOT NULL,
            spec_root_id INTEGER
        );
        ",
    )?;
    ensure_invocations_spec_root_column(&conn)?;
    let spec_root_id = upsert_spec_root(&conn, spec_root)?;
    let ts = unix_ts();
    let argv_json = serde_json::to_string(argv).unwrap_or_else(|_| "[]".to_string());
    let cwd_s = cwd.to_string_lossy();
    conn.execute(
        "INSERT INTO invocations (ts, git_branch, cwd, command_path, argv_json, exit_code, spec_root_id)
         VALUES (?1, ?2, ?3, ?4, ?5, ?6, ?7)",
        rusqlite::params![
            ts,
            branch,
            cwd_s.as_ref(),
            command_path,
            argv_json,
            exit_code,
            spec_root_id
        ],
    )?;
    Ok(())
}

fn unix_ts() -> String {
    use std::time::SystemTime;
    SystemTime::now()
        .duration_since(std::time::UNIX_EPOCH)
        .unwrap_or_default()
        .as_secs()
        .to_string()
}

#[derive(Debug)]
pub struct MatchOutcome<'a> {
    pub chain: Vec<String>,
    pub node: Option<&'a CommandNode>,
    pub trailing: Vec<OsString>,
    pub wants_help: bool,
}

pub fn match_commands<'a>(spec: &'a RootSpec, args: &[OsString]) -> MatchOutcome<'a> {
    let mut chain = Vec::new();
    let mut node: Option<&'a CommandNode> = None;
    let mut map: &BTreeMap<String, CommandNode> = &spec.commands;
    let mut i = 0usize;
    let len = args.len();
    while i < len {
        let raw = &args[i];
        if raw == "--help" || raw == "-h" {
            return MatchOutcome {
                chain,
                node,
                trailing: args[i + 1..].to_vec(),
                wants_help: true,
            };
        }
        let key = raw.to_string_lossy();
        if let Some(next) = map.get(key.as_ref()) {
            chain.push(key.into_owned());
            node = Some(next);
            map = &next.commands;
            i += 1;
            continue;
        }
        break;
    }
    MatchOutcome {
        chain,
        node,
        trailing: args[i..].to_vec(),
        wants_help: false,
    }
}

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

    #[test]
    fn examples_default_spec_validates() {
        let path = PathBuf::from(env!("CARGO_MANIFEST_DIR")).join("examples/default.spec.yaml");
        load_spec(&path).unwrap();
    }

    #[test]
    fn merge_specs_adds_and_replaces_leaves() {
        let mut base = load_spec_from_str(
            r"
commands:
  a:
    about: base
    commands:
      x:
        about: old
        exec:
          argv: [echo, old]
",
            None,
        )
        .unwrap();
        let overlay = load_spec_from_str(
            r"
commands:
  a:
    commands:
      x:
        about: new leaf
        exec:
          argv: [echo, new]
  b:
    about: added top
    exec:
      argv: [echo, b]
",
            None,
        )
        .unwrap();
        merge_specs_into(&mut base, overlay).unwrap();
        base.commands["a"].commands["x"].validate("a x").unwrap();
        assert_eq!(
            base.commands["a"].commands["x"].exec.as_ref().unwrap().argv,
            vec!["echo", "new"]
        );
        assert_eq!(
            base.commands["b"].exec.as_ref().unwrap().argv,
            vec!["echo", "b"]
        );
    }

    #[test]
    fn validate_rejects_exec_with_children() {
        let mut tmp = tempfile::NamedTempFile::with_suffix(".yaml").unwrap();
        write!(
            tmp,
            r"
commands:
  x:
    exec:
      argv: [echo]
    commands:
      child:
        about: nested
"
        )
        .unwrap();
        let err = load_spec(tmp.path()).unwrap_err();
        assert!(err.to_string().contains("cannot define both"));
    }

    #[test]
    fn shell_inline_c_needs_argv0_detects_bash_lc() {
        let argv = vec![
            "bash".into(),
            "-lc".into(),
            "case \"$1\" in create) ;; esac".into(),
        ];
        assert!(shell_inline_c_needs_argv0(&argv));
        let with_placeholder = vec!["zsh".into(), "-c".into(), "echo".into(), "issue".into()];
        assert!(!shell_inline_c_needs_argv0(&with_placeholder));
        assert!(!shell_inline_c_needs_argv0(&[
            "echo".into(),
            "start".into()
        ]));
        assert!(!shell_inline_c_needs_argv0(&[
            "python3".into(),
            "-c".into(),
            "print(1)".into()
        ]));
    }

    #[test]
    fn shell_passthrough_argv0_skips_run_leaf() {
        assert_eq!(
            shell_passthrough_argv0(&[
                "scripts".into(),
                "misc".into(),
                "issue".into(),
                "run".into()
            ]),
            "issue"
        );
        assert_eq!(shell_passthrough_argv0(&["probe".into()]), "probe");
    }

    #[test]
    fn language_exec_path_vs_inline_detection() {
        assert!(ExecSpec::python_value_is_path("scripts/x.py"));
        assert!(ExecSpec::python_value_is_path("X.PY"));
        assert!(!ExecSpec::python_value_is_path("print(1)\n"));
        assert!(!ExecSpec::python_value_is_path("import sys"));
        assert!(ExecSpec::node_value_is_path("a.js"));
        assert!(ExecSpec::node_value_is_path("a.mjs"));
        assert!(ExecSpec::node_value_is_path("a.cjs"));
        assert!(!ExecSpec::node_value_is_path("console.log(1)"));
        assert!(!ExecSpec::node_value_is_path("x.ts"));
        assert!(ExecSpec::bash_value_is_path("x.sh"));
        assert!(ExecSpec::bash_value_is_path("x.bash"));
        assert!(!ExecSpec::bash_value_is_path("echo hi"));
        assert!(ExecSpec::sh_value_is_path("x.sh"));
        assert!(ExecSpec::zsh_value_is_path("x.zsh"));
        let bash = ExecSpec {
            bash: Some("echo hi".into()),
            ..Default::default()
        };
        bash.validate("t").unwrap();

        let python = ExecSpec {
            python: Some("print(1)".into()),
            ..Default::default()
        };
        python.validate("t").unwrap();
        let node = ExecSpec {
            node: Some("console.log(1)".into()),
            ..Default::default()
        };
        node.validate("t").unwrap();
        let both = ExecSpec {
            python: Some("x.py".into()),
            node: Some("x.js".into()),
            ..Default::default()
        };
        assert!(both.validate("t").is_err());
        let text = ExecSpec {
            text: Some("hello docs\n".into()),
            ..Default::default()
        };
        text.validate("t").unwrap();
        let cat: ExecSpec = serde_yaml::from_str("cat: |\n  printed as-is\n").unwrap();
        assert_eq!(cat.literal_text(), Some("printed as-is"));
    }

    #[test]
    fn format_help_inlines_help_child_and_hides_help_leaf() {
        let spec = load_spec_from_str(
            r#"
commands:
  backup:
    about: Backup a path
    inputs:
      path:
        required: true
        type: path
    commands:
      help:
        about: Describe this script.
        exec:
          text: |
            backup — copy files
            Example: jan backup run --path /data
      run:
        about: Run the backup
        exec:
          argv: [echo, ok]
"#,
            None,
        )
        .unwrap();
        let node = &spec.commands["backup"];
        let help = format_help(&spec, &["backup".into()], Some(node));
        assert!(help.contains("backup — copy files"));
        assert!(help.contains("jan backup run --path /data"));
        assert!(help.contains("  run — Run the backup"));
        assert!(!help.contains("  help —"));
        assert!(help.contains("--path"));
        let run = &node.commands["run"];
        let run_help = format_help(&spec, &["backup".into(), "run".into()], Some(run));
        assert!(run_help.contains("backup — copy files"));
        assert!(run_help.contains("--path"));
    }

    #[test]
    fn gherkin_test_names() {
        assert!(gherkin_test_name(
            "given_a_csv_when_summarized_then_prints_shape"
        ));
        assert!(gherkin_test_name(
            "given a file when basename then prints name"
        ));
        assert!(gherkin_test_name(
            "given-a-name-when-run-then-mentions-birthday"
        ));
        assert!(!gherkin_test_name("prints_hello"));
        assert!(!gherkin_test_name("given_when_then"));
        assert!(!gherkin_test_name("given_x_when_y"));
        let t = CommandTest {
            when: "jan hello".into(),
            then: "test \"$JAN_STATUS\" -eq 0".into(),
            ..Default::default()
        };
        t.validate("hello", "given_no_args_when_run_then_ok")
            .unwrap();
        assert!(t.validate("hello", "not_gherkin").is_err());
    }
}

pub fn default_db_path() -> PathBuf {
    if let Ok(p) = std::env::var("JAN_DB") {
        return PathBuf::from(p);
    }
    dirs::data_local_dir()
        .unwrap_or_else(|| PathBuf::from("."))
        .join("jan-cli")
        .join("audit.db")
}