jan-cli 0.13.0

//! Script dependency paths and external utility checks for `exec` leaves.

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

use anyhow::{bail, Context, Result};

use crate::{CommandNode, EnvSpec, RootSpec, RunContext};

#[derive(Debug, Clone)]
struct ScriptEntry {
    path: PathBuf,
    dependencies: Vec<String>,
}

/// Host variables always copied into a restricted child environment when present.
const ESSENTIAL_ENV: &[&str] = &[
    "PATH",
    "HOME",
    "USER",
    "LOGNAME",
    "SHELL",
    "LANG",
    "LC_ALL",
    "LC_CTYPE",
    "LC_MESSAGES",
    "TERM",
    "TERMINFO",
    "COLORTERM",
    "NO_COLOR",
    "TMPDIR",
    "TMP",
    "TEMP",
    "XDG_RUNTIME_DIR",
    "XDG_CONFIG_HOME",
    "XDG_DATA_HOME",
    "XDG_CACHE_HOME",
    "XDG_STATE_HOME",
];

/// Merge `env` from nodes along `chain`; later segments override earlier public keys
/// and pass ids, and append private names (deduped, order preserved).
pub fn collect_chain_env(chain: &[String], spec: &RootSpec) -> EnvSpec {
    let mut env = EnvSpec::default();
    let mut map = &spec.commands;

    for seg in chain {
        let Some(node) = map.get(seg) else { break };
        env.merge_from(node.env.clone());
        map = &node.commands;
    }
    env
}

/// Fail if any private env name is unset in jan's process environment.
pub fn check_private_env(names: &[String]) -> Result<()> {
    let mut missing = Vec::new();
    for name in names {
        let name = name.trim();
        if name.is_empty() {
            bail!("private env entry must not be empty");
        }
        if std::env::var_os(name).is_none() {
            missing.push(name.to_string());
        }
    }
    if missing.is_empty() {
        return Ok(());
    }
    bail!(
        "missing required environment variables: {} \
         (declare them as `env.private` and export them in the host shell; values are not read from YAML)",
        missing.join(", ")
    );
}

/// Fetch a password from the `pass` CLI. Uses the first line of stdout (standard
/// for multiline store entries that keep the secret on line 1).
pub fn fetch_pass_secret(pass_id: &str) -> Result<String> {
    let pass_id = pass_id.trim();
    if pass_id.is_empty() {
        bail!("pass id must not be empty");
    }
    if pass_id.starts_with('-')
        || pass_id.contains('\0')
        || pass_id.contains('\n')
        || pass_id.contains('\r')
    {
        bail!("invalid pass id `{pass_id}`");
    }
    let output = Command::new("pass")
        .arg(pass_id)
        .output()
        .with_context(|| format!("spawn `pass {pass_id}` (is `pass` installed?)"))?;
    if !output.status.success() {
        let stderr = String::from_utf8_lossy(&output.stderr);
        let detail = stderr.trim();
        if detail.is_empty() {
            bail!("`pass {pass_id}` failed with status {}", output.status);
        }
        bail!("`pass {pass_id}` failed: {detail}");
    }
    let text = String::from_utf8(output.stdout)
        .context("`pass` output was not valid UTF-8")?;
    let secret = text
        .lines()
        .next()
        .unwrap_or("")
        .trim_end_matches(['\r', '\n'])
        .to_string();
    if secret.is_empty() {
        bail!("`pass {pass_id}` returned an empty password");
    }
    Ok(secret)
}

/// Resolve all `env.pass` entries into env-var → secret pairs.
pub fn resolve_pass_env(pass: &BTreeMap<String, String>) -> Result<BTreeMap<String, String>> {
    let mut out = BTreeMap::new();
    for (env_name, pass_id) in pass {
        let env_name = env_name.trim();
        if env_name.is_empty() {
            bail!("env.pass key must not be empty");
        }
        let secret = fetch_pass_secret(pass_id)?;
        out.insert(env_name.to_string(), secret);
    }
    Ok(out)
}

/// Apply env policy to `cmd`: either overlay public/PATH onto the inherited env, or
/// clear and rebuild from essentials + private copies + pass secrets + public assignments.
pub fn apply_process_env(
    cmd: &mut Command,
    env: &EnvSpec,
    path_override: Option<String>,
) -> Result<()> {
    if env.restricts_child_env() {
        let built = build_restricted_env(env, path_override)?;
        cmd.env_clear();
        for (key, value) in built {
            cmd.env(key, value);
        }
    } else if let Some(path) = path_override {
        cmd.env("PATH", path);
    }
    Ok(())
}

fn build_restricted_env(
    env: &EnvSpec,
    path_override: Option<String>,
) -> Result<BTreeMap<String, String>> {
    let mut out = BTreeMap::new();
    for key in ESSENTIAL_ENV {
        if let Ok(v) = std::env::var(key) {
            out.insert((*key).to_string(), v);
        }
    }
    for name in &env.private {
        let name = name.trim();
        let value = std::env::var(name)
            .with_context(|| format!("read private env `{name}`"))?;
        out.insert(name.to_string(), value);
    }
    for (k, v) in &env.public {
        out.insert(k.clone(), v.clone());
    }
    // Pass secrets are applied last so they win over public assignments.
    for (k, v) in resolve_pass_env(&env.pass)? {
        out.insert(k, v);
    }
    if let Some(path) = path_override {
        out.insert("PATH".into(), path);
    }
    Ok(out)
}

/// Collect `dependencies`, `requires`, and the nearest `path` from nodes along `chain`.
pub fn collect_chain_metadata(
    chain: &[String],
    spec: &RootSpec,
) -> (Vec<String>, Vec<String>, Option<String>) {
    let mut deps = Vec::new();
    let mut requires = Vec::new();
    let mut path = None;
    let mut map = &spec.commands;

    for seg in chain {
        let Some(node) = map.get(seg) else { break };
        deps.extend(node.dependencies.iter().cloned());
        requires.extend(node.requires.iter().cloned());
        if node.path.is_some() {
            path = node.path.clone();
        }
        map = &node.commands;
    }

    deps.sort();
    deps.dedup();
    requires.sort();
    requires.dedup();
    (deps, requires, path)
}

fn resolve_path_str(raw: &str, ctx: &RunContext<'_>) -> Result<PathBuf> {
    let p = Path::new(raw.trim());
    if p.is_absolute() {
        return p
            .canonicalize()
            .with_context(|| format!("resolve path {}", p.display()));
    }
    if let Ok(root) = std::env::var("JAN_SCRIPTS_ROOT") {
        let candidate = PathBuf::from(root.trim()).join(p);
        if candidate.is_dir() {
            return candidate
                .canonicalize()
                .with_context(|| format!("resolve path {}", candidate.display()));
        }
    }
    let spec_dir = PathBuf::from(&ctx.spec_root.spec_dir);
    for base in [
        ctx.cwd,
        spec_dir.as_path(),
        spec_dir.parent().unwrap_or(Path::new(".")),
    ] {
        let candidate = base.join(p);
        if candidate.is_dir() {
            return candidate
                .canonicalize()
                .with_context(|| format!("resolve path {}", candidate.display()));
        }
    }
    bail!(
        "could not resolve script path `{}` (tried cwd, spec dir, spec parent, and JAN_SCRIPTS_ROOT)",
        raw
    );
}

fn index_scripts(spec: &RootSpec, ctx: &RunContext<'_>) -> Result<BTreeMap<String, ScriptEntry>> {
    let mut index = BTreeMap::new();
    index_commands(&spec.commands, ctx, &mut index)?;
    Ok(index)
}

fn index_commands(
    map: &BTreeMap<String, CommandNode>,
    ctx: &RunContext<'_>,
    index: &mut BTreeMap<String, ScriptEntry>,
) -> Result<()> {
    for (name, node) in map {
        if let Some(ref raw_path) = node.path {
            if let Ok(path) = resolve_path_str(raw_path, ctx) {
                let entry = ScriptEntry {
                    path,
                    dependencies: node.dependencies.clone(),
                };
                if let Some(prev) = index.get(name) {
                    if prev.path != entry.path {
                        bail!(
                            "duplicate script name `{name}` with different paths ({} vs {})",
                            prev.path.display(),
                            entry.path.display()
                        );
                    }
                }
                index.insert(name.clone(), entry);
            }
        }
        index_commands(&node.commands, ctx, index)?;
    }
    Ok(())
}

fn visit_dependency(
    name: &str,
    index: &BTreeMap<String, ScriptEntry>,
    visiting: &mut HashSet<String>,
    visited: &mut HashSet<String>,
    ordered: &mut Vec<PathBuf>,
) -> Result<()> {
    if visited.contains(name) {
        return Ok(());
    }
    if !visiting.insert(name.to_string()) {
        bail!("cyclic script dependency involving `{name}`");
    }
    let Some(entry) = index.get(name) else {
        // Dependency may be satisfied by inlined `run` wrappers when source dirs are absent.
        return Ok(());
    };
    for dep in &entry.dependencies {
        visit_dependency(dep, index, visiting, visited, ordered)?;
    }
    visiting.remove(name);
    visited.insert(name.to_string());
    ordered.push(entry.path.clone());
    Ok(())
}

/// Resolve transitive script dependency directories, then the invoking script's own `path`.
pub fn resolve_path_prefixes(
    spec: &RootSpec,
    chain: &[String],
    ctx: &RunContext<'_>,
) -> Result<Vec<PathBuf>> {
    let (dep_names, _, own_path) = collect_chain_metadata(chain, spec);
    let index = index_scripts(spec, ctx)?;
    let mut visiting = HashSet::new();
    let mut visited = HashSet::new();
    let mut dirs = Vec::new();

    for name in &dep_names {
        visit_dependency(name, &index, &mut visiting, &mut visited, &mut dirs)?;
    }
    if let Some(raw) = own_path {
        if let Ok(own) = resolve_path_str(&raw, ctx) {
            if !dirs.iter().any(|p| p == &own) {
                dirs.push(own);
            }
        }
    }
    Ok(dirs)
}

pub fn prepend_path_env(dirs: &[PathBuf]) -> Result<String> {
    let current = std::env::var("PATH").unwrap_or_default();
    let sep = if cfg!(windows) { ";" } else { ":" };
    let mut parts: Vec<String> = dirs
        .iter()
        .map(|p| p.to_string_lossy().into_owned())
        .collect();
    if !current.is_empty() {
        parts.push(current);
    }
    Ok(parts.join(sep))
}

/// Ensure each named utility is discoverable on PATH before running the leaf.
pub fn check_requires(requires: &[String]) -> Result<()> {
    let mut missing = Vec::new();
    for name in requires {
        let name = name.trim();
        if name.is_empty() {
            continue;
        }
        if !utility_available(name) {
            missing.push(name.to_string());
        }
    }
    if missing.is_empty() {
        return Ok(());
    }
    bail!(
        "missing required utilities on PATH: {} (install them or adjust the spec `requires` list)",
        missing.join(", ")
    );
}

/// Directories on the PATH jan itself inherited, before any spec `path:` or
/// `dependencies` entries are prepended.
fn inherited_path_dirs() -> Vec<PathBuf> {
    let Ok(path_var) = std::env::var("PATH") else {
        return Vec::new();
    };
    let sep = if cfg!(windows) { ';' } else { ':' };
    path_var
        .split(sep)
        .filter(|d| !d.is_empty())
        .map(PathBuf::from)
        .collect()
}

/// Absolute program to spawn for `exec.argv[0]`.
///
/// Resolution prefers the PATH jan inherited, which is also the PATH that
/// `requires` was checked against. A spec therefore cannot use `path:` or
/// `dependencies` to substitute its own `bash`/`python3` for the system one.
/// Spec directories are consulted only for programs the host does not provide,
/// where no system binary is being shadowed.
pub fn resolve_program(program: &str, spec_dirs: &[PathBuf]) -> Result<PathBuf> {
    resolve_program_in(program, &inherited_path_dirs(), spec_dirs)
}

fn resolve_program_in(
    program: &str,
    inherited_dirs: &[PathBuf],
    spec_dirs: &[PathBuf],
) -> Result<PathBuf> {
    let trimmed = program.trim();
    if trimmed.is_empty() {
        bail!("exec.argv[0] must not be empty");
    }
    let candidate = Path::new(trimmed);
    // An explicit path is the spec author's own choice, not a PATH lookup.
    if candidate.components().count() > 1 {
        return Ok(candidate.to_path_buf());
    }
    if let Some(found) = lookup_in_dirs(trimmed, inherited_dirs) {
        return Ok(found);
    }
    if let Some(found) = lookup_in_dirs(trimmed, spec_dirs) {
        return Ok(found);
    }
    bail!("program `{trimmed}` not found on PATH (from exec.argv)");
}

fn utility_available(name: &str) -> bool {
    if which_in_path(name) {
        return true;
    }
    // Windows may need PATHEXT; `where` handles that.
    Command::new(if cfg!(windows) { "where" } else { "which" })
        .arg(name)
        .stdout(std::process::Stdio::null())
        .stderr(std::process::Stdio::null())
        .status()
        .map(|s| s.success())
        .unwrap_or(false)
}

fn which_in_path(name: &str) -> bool {
    lookup_in_dirs(name, &inherited_path_dirs()).is_some()
}

fn lookup_in_dirs(name: &str, dirs: &[PathBuf]) -> Option<PathBuf> {
    for dir in dirs {
        let candidate = dir.join(name);
        if candidate.is_file() {
            return Some(candidate);
        }
        #[cfg(windows)]
        {
            for ext in ["exe", "cmd", "bat", "com"] {
                let with_ext = dir.join(format!("{name}.{ext}"));
                if with_ext.is_file() {
                    return Some(with_ext);
                }
            }
        }
    }
    None
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::{EnvSpec, ExecSpec, SpecRootIdentity};
    use std::collections::BTreeMap;
    use std::fs;
    use tempfile::tempdir;

    #[test]
    fn transitive_dependencies_prepended_in_order() {
        let dir = tempdir().unwrap();
        let dep_a = dir.path().join("a");
        let dep_b = dir.path().join("b");
        let main = dir.path().join("main");
        fs::create_dir_all(&dep_a).unwrap();
        fs::create_dir_all(&dep_b).unwrap();
        fs::create_dir_all(&main).unwrap();

        let spec = RootSpec {
            metadata: None,
            commands: BTreeMap::from([
                (
                    "a".into(),
                    CommandNode {
                        path: Some(dep_a.to_string_lossy().into_owned()),
                        dependencies: vec![],
                        ..Default::default()
                    },
                ),
                (
                    "b".into(),
                    CommandNode {
                        path: Some(dep_b.to_string_lossy().into_owned()),
                        dependencies: vec!["a".into()],
                        ..Default::default()
                    },
                ),
                (
                    "main".into(),
                    CommandNode {
                        path: Some(main.to_string_lossy().into_owned()),
                        dependencies: vec!["b".into()],
                        commands: BTreeMap::from([(
                            "run".into(),
                            CommandNode {
                                exec: Some(ExecSpec {
                                    argv: vec!["echo".into()],
                                    passthrough: false,
                                    ..Default::default()
                                }),
                                ..Default::default()
                            },
                        )]),
                        ..Default::default()
                    },
                ),
            ]),
        };

        let identity = SpecRootIdentity {
            spec_dir: dir.path().to_string_lossy().into_owned(),
            root_yaml: "spec.yaml".into(),
        };
        let ctx = RunContext {
            cwd: dir.path(),
            db_path: None,
            branch: "test".into(),
            no_log: true,
            spec_root: &identity,
        };
        let dirs = resolve_path_prefixes(&spec, &["main".into(), "run".into()], &ctx).unwrap();
        assert_eq!(dirs.len(), 3);
        assert_eq!(dirs[0], dep_a.canonicalize().unwrap());
        assert_eq!(dirs[1], dep_b.canonicalize().unwrap());
        assert_eq!(dirs[2], main.canonicalize().unwrap());
    }

    #[test]
    fn cyclic_dependency_errors() {
        let dir = tempdir().unwrap();
        let a = dir.path().join("a");
        let b = dir.path().join("b");
        fs::create_dir_all(&a).unwrap();
        fs::create_dir_all(&b).unwrap();

        let spec = RootSpec {
            metadata: None,
            commands: BTreeMap::from([
                (
                    "a".into(),
                    CommandNode {
                        path: Some(a.to_string_lossy().into_owned()),
                        dependencies: vec!["b".into()],
                        ..Default::default()
                    },
                ),
                (
                    "b".into(),
                    CommandNode {
                        path: Some(b.to_string_lossy().into_owned()),
                        dependencies: vec!["a".into()],
                        ..Default::default()
                    },
                ),
                (
                    "run".into(),
                    CommandNode {
                        dependencies: vec!["a".into()],
                        exec: Some(ExecSpec {
                            argv: vec!["echo".into()],
                            passthrough: false,
                            ..Default::default()
                        }),
                        ..Default::default()
                    },
                ),
            ]),
        };

        let identity = SpecRootIdentity {
            spec_dir: dir.path().to_string_lossy().into_owned(),
            root_yaml: "spec.yaml".into(),
        };
        let ctx = RunContext {
            cwd: dir.path(),
            db_path: None,
            branch: "test".into(),
            no_log: true,
            spec_root: &identity,
        };
        let err = resolve_path_prefixes(&spec, &["run".into()], &ctx).unwrap_err();
        assert!(err.to_string().contains("cyclic"));
    }

    #[test]
    fn chain_env_later_overrides_earlier() {
        let spec = RootSpec {
            metadata: None,
            commands: BTreeMap::from([(
                "a".into(),
                CommandNode {
                    env: EnvSpec {
                        public: BTreeMap::from([
                            ("X".into(), "1".into()),
                            ("Y".into(), "a".into()),
                        ]),
                        private: vec!["SECRET".into()],
                        pass: BTreeMap::from([("TOKEN".into(), "github/pat".into())]),
                    },
                    commands: BTreeMap::from([(
                        "b".into(),
                        CommandNode {
                            env: EnvSpec {
                                public: BTreeMap::from([("X".into(), "2".into())]),
                                private: vec!["OTHER".into()],
                                pass: BTreeMap::from([("TOKEN".into(), "github/other".into())]),
                            },
                            ..Default::default()
                        },
                    )]),
                    ..Default::default()
                },
            )]),
        };
        let env = collect_chain_env(&["a".into(), "b".into()], &spec);
        assert_eq!(env.public.get("X").map(String::as_str), Some("2"));
        assert_eq!(env.public.get("Y").map(String::as_str), Some("a"));
        assert_eq!(env.private, vec!["SECRET".to_string(), "OTHER".to_string()]);
        assert_eq!(
            env.pass.get("TOKEN").map(String::as_str),
            Some("github/other")
        );
    }

    #[test]
    fn env_pass_conflicts_with_private() {
        let env = EnvSpec {
            private: vec!["GH_TOKEN".into()],
            pass: BTreeMap::from([("GH_TOKEN".into(), "github/pat".into())]),
            ..Default::default()
        };
        let err = env.validate("issue").unwrap_err();
        assert!(err.to_string().contains("both `env.private` and `env.pass`"));
    }

    #[test]
    fn fetch_pass_uses_fake_pass_on_path() {
        let dir = tempfile::tempdir().unwrap();
        let fake = dir.path().join("pass");
        std::fs::write(
            &fake,
            "#!/bin/sh\ncase \"$1\" in\n  github/pat) printf 's3cret\\nnote line\\n';;\n  *) exit 1;;\nesac\n",
        )
        .unwrap();
        #[cfg(unix)]
        {
            use std::os::unix::fs::PermissionsExt;
            let mut perms = std::fs::metadata(&fake).unwrap().permissions();
            perms.set_mode(0o755);
            std::fs::set_permissions(&fake, perms).unwrap();
        }
        let old_path = std::env::var_os("PATH");
        let new_path = format!(
            "{}{}{}",
            dir.path().display(),
            std::path::MAIN_SEPARATOR,
            old_path
                .as_ref()
                .map(|p| p.to_string_lossy().into_owned())
                .unwrap_or_default()
                .trim_start_matches(|c| c == std::path::MAIN_SEPARATOR)
                .is_empty()
                .then(|| String::new())
                .unwrap_or_else(|| format!(
                    "{}{}",
                    if cfg!(windows) { ";" } else { ":" },
                    old_path
                        .as_ref()
                        .map(|p| p.to_string_lossy().into_owned())
                        .unwrap_or_default()
                ))
        );
        // Simpler PATH prepend:
        let sep = if cfg!(windows) { ";" } else { ":" };
        let new_path = match &old_path {
            Some(p) => format!("{}{}{}", dir.path().display(), sep, p.to_string_lossy()),
            None => dir.path().display().to_string(),
        };
        std::env::set_var("PATH", &new_path);
        let secret = fetch_pass_secret("github/pat").unwrap();
        assert_eq!(secret, "s3cret");
        match old_path {
            Some(p) => std::env::set_var("PATH", p),
            None => std::env::remove_var("PATH"),
        }
        let _ = new_path; // silence if unused in windows branch cleanup
    }

    #[test]
    fn check_private_env_reports_missing() {
        let err = check_private_env(&["JAN_TEST_MISSING_ENV_VAR_XYZ".to_string()]).unwrap_err();
        assert!(err.to_string().contains("missing required environment variables"));
    }

    #[test]
    fn check_requires_reports_missing() {
        let err = check_requires(&["definitely-not-a-real-binary-xyz".to_string()]).unwrap_err();
        assert!(err.to_string().contains("missing required utilities"));
    }

    fn touch_executable(dir: &Path, name: &str) -> PathBuf {
        let p = dir.join(name);
        fs::write(&p, "#!/bin/sh\n").unwrap();
        p
    }

    #[test]
    fn spec_path_cannot_shadow_inherited_program() {
        let dir = tempdir().unwrap();
        let system = dir.path().join("system");
        let spec_dir = dir.path().join("spec");
        fs::create_dir_all(&system).unwrap();
        fs::create_dir_all(&spec_dir).unwrap();
        let real = touch_executable(&system, "sh-like");
        touch_executable(&spec_dir, "sh-like");

        let resolved = resolve_program_in(
            "sh-like",
            std::slice::from_ref(&system),
            std::slice::from_ref(&spec_dir),
        )
        .unwrap();
        assert_eq!(resolved, real);
    }

    #[test]
    fn spec_path_supplies_programs_absent_from_inherited_path() {
        let dir = tempdir().unwrap();
        let spec_dir = dir.path().join("spec");
        fs::create_dir_all(&spec_dir).unwrap();
        let helper = touch_executable(&spec_dir, "helper-only");

        let resolved =
            resolve_program_in("helper-only", &[], std::slice::from_ref(&spec_dir)).unwrap();
        assert_eq!(resolved, helper);
    }

    #[test]
    fn unresolvable_program_errors_before_spawn() {
        let err = resolve_program_in("definitely-not-a-real-binary-xyz", &[], &[]).unwrap_err();
        assert!(err.to_string().contains("not found on PATH"));
    }

    #[test]
    fn explicit_paths_are_left_alone() {
        let resolved = resolve_program_in("./tools/build.sh", &[], &[]).unwrap();
        assert_eq!(resolved, PathBuf::from("./tools/build.sh"));
    }
}