kmp_embedded/

data_dir.rs

use std::ffi::OsString;
use std::fs;
use std::path::{Path, PathBuf};

use kmp_domain::PortError;

/// Explicit data directory override (ADR-012 rule 1).
pub const DATA_DIR_ENV: &str = "KMP_MCP_DATA_DIR";

const PROJECT_DIR_NAME: &str = ".kernel";

/// Where a project keeps the committed copy of its memory, relative to the
/// project root.
///
/// The store itself (`.kernel/`) is machine state and is auto-gitignored. A
/// bundle is the event log in one text file, which is a different thing: it
/// belongs to the repository the same way a migration or a fixture does, so
/// memory branches, reviews and reverts with the code that produced it.
///
/// The path is a convention rather than a setting so that `export` and
/// `import` with no argument mean the same thing in every checkout, and so a
/// reviewer knows where to look.
pub const PROJECT_BUNDLE_PATH: &str = ".kmp/memory.jsonl";

/// Where the data directory came from — logged at startup so the winning
/// resolution rule is always visible.
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum ResolvedDataDir {
    /// `KMP_MCP_DATA_DIR` was set.
    Explicit(PathBuf),
    /// `<project-root>/.kernel/`, project root found by walking up to `.git`.
    Project(PathBuf),
    /// Per-user fallback under the platform data dir.
    UserDefault(PathBuf),
}

impl ResolvedDataDir {
    pub fn path(&self) -> &Path {
        match self {
            Self::Explicit(path) | Self::Project(path) | Self::UserDefault(path) => path,
        }
    }

    pub fn rule_name(&self) -> &'static str {
        match self {
            Self::Explicit(_) => "env",
            Self::Project(_) => "project",
            Self::UserDefault(_) => "user",
        }
    }
}

/// ADR-012 resolution: env override > project `.kernel/` > per-user default.
/// Pure function for testability; `resolve_data_dir_from_env` feeds it from
/// the process environment.
pub fn resolve_data_dir(
    env_override: Option<&str>,
    working_dir: &Path,
    user_data_home: &Path,
) -> ResolvedDataDir {
    resolve_with_project_marker(env_override, working_dir, user_data_home, |candidate| {
        candidate.join(".git").exists()
    })
}

fn resolve_with_project_marker(
    env_override: Option<&str>,
    working_dir: &Path,
    user_data_home: &Path,
    is_project_root: impl Fn(&Path) -> bool,
) -> ResolvedDataDir {
    if let Some(explicit) = env_override
        .map(str::trim)
        .filter(|value| !value.is_empty())
    {
        return ResolvedDataDir::Explicit(PathBuf::from(explicit));
    }

    let mut current = Some(working_dir);
    while let Some(candidate) = current {
        if is_project_root(candidate) {
            return ResolvedDataDir::Project(candidate.join(PROJECT_DIR_NAME));
        }
        current = candidate.parent();
    }

    ResolvedDataDir::UserDefault(user_data_home.join("kmp").join("default"))
}

/// Where user-scope memory lives: the Unix data home when available, then
/// the native Windows local-data directories.
///
/// Exposed because it is also where anything that wants to enumerate the
/// machine's memories has to look, and a second copy of this rule would be a
/// second answer to the same question.
pub fn user_data_home() -> Option<PathBuf> {
    user_data_home_from(|name| std::env::var_os(name))
}

fn user_data_home_from(mut read: impl FnMut(&str) -> Option<OsString>) -> Option<PathBuf> {
    let path = |value: Option<OsString>| {
        value
            .map(PathBuf::from)
            .filter(|candidate| !candidate.as_os_str().is_empty())
    };

    path(read("XDG_DATA_HOME"))
        .or_else(|| path(read("HOME")).map(|home| home.join(".local").join("share")))
        .or_else(|| path(read("LOCALAPPDATA")))
        .or_else(|| path(read("APPDATA")))
        .or_else(|| path(read("USERPROFILE")).map(|home| home.join("AppData").join("Local")))
}

/// The conventional bundle path for the project `data_dir` belongs to.
///
/// Only a project-scoped store has one: an explicit `KMP_MCP_DATA_DIR` or the
/// per-user default has no repository to be committed to, and guessing one
/// would put memory somewhere the operator did not choose.
pub fn project_bundle_path(resolved: &ResolvedDataDir) -> Option<PathBuf> {
    match resolved {
        ResolvedDataDir::Project(path) => path
            .parent()
            .map(|project_root| project_root.join(PROJECT_BUNDLE_PATH)),
        ResolvedDataDir::Explicit(_) | ResolvedDataDir::UserDefault(_) => None,
    }
}

/// Resolves from the process environment and prepares the directory. Every
/// data directory gets the same safety skeleton, regardless of whether it was
/// discovered from a project, supplied explicitly, or created by migration.
pub fn resolve_data_dir_from_env() -> Result<ResolvedDataDir, PortError> {
    let resolved = locate_data_dir_from_env()?;
    prepare_data_dir(&resolved)?;
    Ok(resolved)
}

/// Resolves from the process environment and touches nothing.
///
/// Reporting where memory *would* live must not bring it into being:
/// `kmp-mcp info` and `kmp-mcp doctor` run wherever a user happens to be
/// standing, and a diagnostic that leaves a `.kernel/` behind in an unrelated
/// repository has answered a question by changing the answer.
pub fn locate_data_dir_from_env() -> Result<ResolvedDataDir, PortError> {
    let env_override = std::env::var(DATA_DIR_ENV).ok();
    let working_dir = std::env::current_dir().map_err(|error| {
        PortError::Unavailable(format!(
            "embedded kernel could not resolve the working directory: {error}"
        ))
    })?;
    let user_data_home = user_data_home().ok_or_else(|| {
        PortError::Unavailable(
            "embedded kernel could not resolve a user data directory \
             (none of XDG_DATA_HOME, HOME, LOCALAPPDATA, APPDATA, or USERPROFILE is set)"
                .to_string(),
        )
    })?;

    Ok(resolve_data_dir(
        env_override.as_deref(),
        &working_dir,
        &user_data_home,
    ))
}

fn prepare_data_dir(resolved: &ResolvedDataDir) -> Result<(), PortError> {
    ensure_data_dir_skeleton(resolved.path())
}

/// Creates the non-store part of a KMP data directory.
///
/// Fresh startup, `migrate`, and `share-memory` all call this function. The
/// self-ignore file is deliberately installed even for an explicit path: an
/// operator can put such a path inside a repository, and the store must not
/// start appearing in `git status` merely because it arrived through a
/// migration rather than first startup. Existing files are never replaced.
pub fn ensure_data_dir_skeleton(path: &Path) -> Result<(), PortError> {
    fs::create_dir_all(path).map_err(|error| {
        PortError::Unavailable(format!(
            "embedded kernel could not create data dir `{}`: {error}",
            path.display()
        ))
    })?;

    let gitignore = path.join(".gitignore");
    if !gitignore.exists() {
        fs::write(&gitignore, "*\n").map_err(|error| {
            PortError::Unavailable(format!(
                "embedded kernel could not write `{}`: {error}",
                gitignore.display()
            ))
        })?;
    }
    let logs = path.join("logs");
    fs::create_dir_all(&logs).map_err(|error| {
        PortError::Unavailable(format!(
            "embedded kernel could not create log dir `{}`: {error}",
            logs.display()
        ))
    })?;
    Ok(())
}

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

    #[test]
    fn only_a_project_store_has_a_conventional_bundle_path() {
        let project = ResolvedDataDir::Project(PathBuf::from("/repo/.kernel"));
        assert_eq!(
            project_bundle_path(&project),
            Some(PathBuf::from("/repo/.kmp/memory.jsonl")),
            "the bundle sits beside the store's project root, not inside the store"
        );

        // Neither of these belongs to a repository, and picking one for them
        // would write memory somewhere nobody chose.
        assert_eq!(
            project_bundle_path(&ResolvedDataDir::Explicit(PathBuf::from("/tmp/dir"))),
            None
        );
        assert_eq!(
            project_bundle_path(&ResolvedDataDir::UserDefault(PathBuf::from("/home/u/kmp"))),
            None
        );
    }

    #[test]
    fn env_override_wins_over_everything() {
        let resolved = resolve_data_dir(
            Some("/explicit/dir"),
            Path::new("/some/project"),
            Path::new("/home/u/.local/share"),
        );
        assert_eq!(
            resolved,
            ResolvedDataDir::Explicit(PathBuf::from("/explicit/dir"))
        );
        assert_eq!(resolved.rule_name(), "env");
    }

    #[test]
    fn blank_env_override_is_ignored() {
        let resolved = resolve_with_project_marker(
            Some("  "),
            Path::new("/anywhere"),
            Path::new("/data"),
            |_| false,
        );
        assert_eq!(resolved.rule_name(), "user");
    }

    #[test]
    fn project_root_is_found_by_walking_up_to_git() {
        let temp = tempfile::tempdir().expect("tempdir");
        let nested = temp.path().join("workspace").join("src");
        std::fs::create_dir_all(&nested).expect("nested dirs");
        std::fs::create_dir_all(temp.path().join("workspace").join(".git")).expect("git dir");

        let resolved = resolve_data_dir(None, &nested, Path::new("/data"));
        assert_eq!(
            resolved,
            ResolvedDataDir::Project(temp.path().join("workspace").join(".kernel"))
        );
    }

    #[test]
    fn no_project_falls_back_to_user_data_dir() {
        let resolved = resolve_with_project_marker(
            None,
            Path::new("/anywhere/nested"),
            Path::new("/home/u/.local/share"),
            |_| false,
        );
        assert_eq!(
            resolved,
            ResolvedDataDir::UserDefault(PathBuf::from("/home/u/.local/share/kmp/default"))
        );
    }

    #[test]
    fn user_data_home_keeps_unix_precedence_and_supports_native_windows() {
        let unix = user_data_home_from(|name| match name {
            "XDG_DATA_HOME" => Some(OsString::from("/xdg")),
            "HOME" => Some(OsString::from("/home/user")),
            "LOCALAPPDATA" => Some(OsString::from(r"C:\Users\user\AppData\Local")),
            _ => None,
        });
        assert_eq!(unix, Some(PathBuf::from("/xdg")));

        let windows = user_data_home_from(|name| match name {
            "LOCALAPPDATA" => Some(OsString::from(r"C:\Users\user\AppData\Local")),
            "APPDATA" => Some(OsString::from(r"C:\Users\user\AppData\Roaming")),
            _ => None,
        });
        assert_eq!(windows, Some(PathBuf::from(r"C:\Users\user\AppData\Local")));

        let profile = user_data_home_from(|name| match name {
            "USERPROFILE" => Some(OsString::from(r"C:\Users\user")),
            _ => None,
        });
        assert_eq!(
            profile,
            Some(
                PathBuf::from(r"C:\Users\user")
                    .join("AppData")
                    .join("Local")
            )
        );
    }

    #[test]
    fn project_dir_preparation_writes_self_ignoring_gitignore() {
        let temp = tempfile::tempdir().expect("tempdir");
        let kernel_dir = temp.path().join(".kernel");
        let resolved = ResolvedDataDir::Project(kernel_dir.clone());

        prepare_data_dir(&resolved).expect("prepare");

        let gitignore = std::fs::read_to_string(kernel_dir.join(".gitignore")).expect("gitignore");
        assert_eq!(gitignore, "*\n");
        assert!(kernel_dir.join("logs").is_dir());
    }

    #[test]
    fn explicit_and_migrated_dirs_get_the_same_non_destructive_skeleton() {
        let temp = tempfile::tempdir().expect("tempdir");
        let data_dir = temp.path().join("destination");

        ensure_data_dir_skeleton(&data_dir).expect("prepare explicit destination");
        assert_eq!(
            std::fs::read_to_string(data_dir.join(".gitignore")).expect("gitignore"),
            "*\n"
        );
        assert!(data_dir.join("logs").is_dir());

        std::fs::write(data_dir.join(".gitignore"), "keep-me\n").expect("custom ignore");
        ensure_data_dir_skeleton(&data_dir).expect("prepare again");
        assert_eq!(
            std::fs::read_to_string(data_dir.join(".gitignore")).expect("custom gitignore"),
            "keep-me\n",
            "the skeleton never overwrites an operator-owned ignore file"
        );
    }
}