projd_core/

discover.rs

//! Project root discovery for directories that may contain multiple independent
//! projects (different VCS, nested vendored repos, monorepos, documentation
//! sites, datasets, ...).
//!
//! `discover_roots` walks a directory tree and classifies each directory as
//! one of `Root` (one or more recognized project root kinds), `Container`
//! (a parent directory worth descending into), or `Skip` (build output,
//! dependency caches, hidden tooling directories). Termination is driven by
//! root recognition rather than depth; `max_depth` only guards against
//! pathological filesystem structures.

use std::collections::{BTreeMap, BTreeSet};
use std::ffi::OsStr;
use std::fs;
use std::path::{Path, PathBuf};

use anyhow::{Context, Result, bail};
use serde::{Deserialize, Serialize};

#[derive(Clone, Copy, Debug, Eq, PartialEq, Hash, Ord, PartialOrd, Serialize, Deserialize)]
#[serde(rename_all = "snake_case")]
pub enum RootKind {
    GitRepo,
    HgRepo,
    SvnRepo,
    FossilRepo,
    BzrRepo,
    CargoWorkspace,
    CargoPackage,
    NpmPackage,
    PnpmWorkspace,
    YarnWorkspace,
    LernaWorkspace,
    TurboWorkspace,
    PythonProject,
    GoWorkspace,
    GoModule,
    MavenProject,
    GradleWorkspace,
    GradleProject,
    CMakeProject,
    RubyProject,
    PhpProject,
    ElixirProject,
    DotnetProject,
    MdBook,
    MkDocs,
    Jekyll,
    Sphinx,
    Docusaurus,
    Hugo,
    Gatsby,
    Astro,
    DocFx,
    DvcDataset,
}

#[derive(Clone, Copy, Debug, Eq, PartialEq, Hash, Ord, PartialOrd, Serialize, Deserialize)]
#[serde(rename_all = "snake_case")]
pub enum RootCategory {
    Vcs,
    Workspace,
    Package,
    Docs,
    Data,
}

#[derive(Clone, Copy, Debug, Eq, PartialEq, Hash, Ord, PartialOrd, Serialize, Deserialize)]
#[serde(rename_all = "snake_case")]
pub enum Confidence {
    Weak,
    Medium,
    Strong,
}

#[derive(Clone, Debug, Eq, PartialEq, Serialize, Deserialize)]
pub struct DiscoveredRoot {
    pub path: PathBuf,
    pub kinds: Vec<RootKind>,
    pub confidence: Confidence,
}

#[derive(Clone, Debug)]
pub struct DiscoverOptions {
    pub max_depth: usize,
    pub min_confidence: Confidence,
    pub include_kinds: Option<BTreeSet<RootKind>>,
    pub expand_workspaces: bool,
    pub nested_vcs: bool,
}

impl Default for DiscoverOptions {
    fn default() -> Self {
        Self {
            max_depth: 8,
            min_confidence: Confidence::Medium,
            include_kinds: None,
            expand_workspaces: false,
            nested_vcs: false,
        }
    }
}

impl RootKind {
    pub fn category(self) -> RootCategory {
        use RootKind::*;
        match self {
            GitRepo | HgRepo | SvnRepo | FossilRepo | BzrRepo => RootCategory::Vcs,
            CargoWorkspace | PnpmWorkspace | YarnWorkspace | LernaWorkspace | TurboWorkspace
            | GoWorkspace | GradleWorkspace => RootCategory::Workspace,
            CargoPackage | NpmPackage | PythonProject | GoModule | MavenProject | GradleProject
            | CMakeProject | RubyProject | PhpProject | ElixirProject | DotnetProject => {
                RootCategory::Package
            }
            MdBook | MkDocs | Jekyll | Sphinx | Docusaurus | Hugo | Gatsby | Astro | DocFx => {
                RootCategory::Docs
            }
            DvcDataset => RootCategory::Data,
        }
    }

    pub fn token(self) -> &'static str {
        use RootKind::*;
        match self {
            GitRepo => "git_repo",
            HgRepo => "hg_repo",
            SvnRepo => "svn_repo",
            FossilRepo => "fossil_repo",
            BzrRepo => "bzr_repo",
            CargoWorkspace => "cargo_workspace",
            CargoPackage => "cargo_package",
            NpmPackage => "npm_package",
            PnpmWorkspace => "pnpm_workspace",
            YarnWorkspace => "yarn_workspace",
            LernaWorkspace => "lerna_workspace",
            TurboWorkspace => "turbo_workspace",
            PythonProject => "python_project",
            GoWorkspace => "go_workspace",
            GoModule => "go_module",
            MavenProject => "maven_project",
            GradleWorkspace => "gradle_workspace",
            GradleProject => "gradle_project",
            CMakeProject => "cmake_project",
            RubyProject => "ruby_project",
            PhpProject => "php_project",
            ElixirProject => "elixir_project",
            DotnetProject => "dotnet_project",
            MdBook => "mdbook",
            MkDocs => "mkdocs",
            Jekyll => "jekyll",
            Sphinx => "sphinx",
            Docusaurus => "docusaurus",
            Hugo => "hugo",
            Gatsby => "gatsby",
            Astro => "astro",
            DocFx => "docfx",
            DvcDataset => "dvc_dataset",
        }
    }

    pub fn from_token(token: &str) -> Option<Self> {
        let normalized = token.trim().to_ascii_lowercase().replace('-', "_");
        match normalized.as_str() {
            "git" | "git_repo" => Some(Self::GitRepo),
            "hg" | "mercurial" | "hg_repo" => Some(Self::HgRepo),
            "svn" | "subversion" | "svn_repo" => Some(Self::SvnRepo),
            "fossil" | "fossil_repo" => Some(Self::FossilRepo),
            "bzr" | "bazaar" | "bzr_repo" => Some(Self::BzrRepo),
            "cargo" | "cargo_package" => Some(Self::CargoPackage),
            "cargo_workspace" => Some(Self::CargoWorkspace),
            "npm" | "npm_package" | "node" => Some(Self::NpmPackage),
            "pnpm" | "pnpm_workspace" => Some(Self::PnpmWorkspace),
            "yarn" | "yarn_workspace" => Some(Self::YarnWorkspace),
            "lerna" | "lerna_workspace" => Some(Self::LernaWorkspace),
            "turbo" | "turbo_workspace" => Some(Self::TurboWorkspace),
            "python" | "pyproject" | "python_project" => Some(Self::PythonProject),
            "go" | "go_module" => Some(Self::GoModule),
            "go_workspace" => Some(Self::GoWorkspace),
            "maven" | "maven_project" => Some(Self::MavenProject),
            "gradle" | "gradle_project" => Some(Self::GradleProject),
            "gradle_workspace" => Some(Self::GradleWorkspace),
            "cmake" | "cmake_project" => Some(Self::CMakeProject),
            "ruby" | "ruby_project" => Some(Self::RubyProject),
            "php" | "composer" | "php_project" => Some(Self::PhpProject),
            "elixir" | "mix" | "elixir_project" => Some(Self::ElixirProject),
            "dotnet" | "csharp" | "dotnet_project" => Some(Self::DotnetProject),
            "mdbook" => Some(Self::MdBook),
            "mkdocs" => Some(Self::MkDocs),
            "jekyll" => Some(Self::Jekyll),
            "sphinx" => Some(Self::Sphinx),
            "docusaurus" => Some(Self::Docusaurus),
            "hugo" => Some(Self::Hugo),
            "gatsby" => Some(Self::Gatsby),
            "astro" => Some(Self::Astro),
            "docfx" => Some(Self::DocFx),
            "dvc" | "dvc_dataset" => Some(Self::DvcDataset),
            _ => None,
        }
    }
}

impl Confidence {
    pub fn token(self) -> &'static str {
        match self {
            Self::Weak => "weak",
            Self::Medium => "medium",
            Self::Strong => "strong",
        }
    }

    pub fn from_token(token: &str) -> Option<Self> {
        match token.trim().to_ascii_lowercase().as_str() {
            "weak" => Some(Self::Weak),
            "medium" => Some(Self::Medium),
            "strong" => Some(Self::Strong),
            _ => None,
        }
    }
}

pub fn discover_roots(
    path: impl AsRef<Path>,
    opts: &DiscoverOptions,
) -> Result<Vec<DiscoveredRoot>> {
    let input = path.as_ref();
    let canonical = fs::canonicalize(input)
        .with_context(|| format!("failed to resolve `{}`", input.display()))?;
    if !canonical.is_dir() {
        bail!("`{}` is not a directory", canonical.display());
    }

    let mut results: Vec<DiscoveredRoot> = Vec::new();
    let mut stack: Vec<(PathBuf, usize)> = vec![(canonical, 0)];

    while let Some((dir, depth)) = stack.pop() {
        match classify_dir(&dir) {
            DirClass::Skip => continue,
            DirClass::Root { kinds, confidence } => {
                let is_workspace = kinds
                    .iter()
                    .any(|kind| kind.category() == RootCategory::Workspace);
                let is_vcs = kinds
                    .iter()
                    .any(|kind| kind.category() == RootCategory::Vcs);
                results.push(DiscoveredRoot {
                    path: dir.clone(),
                    kinds,
                    confidence,
                });

                let descend =
                    (is_workspace && opts.expand_workspaces) || (is_vcs && opts.nested_vcs);
                if !descend || depth >= opts.max_depth {
                    continue;
                }
                push_children(&dir, depth, &mut stack);
            }
            DirClass::Container => {
                if depth >= opts.max_depth {
                    continue;
                }
                push_children(&dir, depth, &mut stack);
            }
        }
    }

    results.retain(|root| root.confidence >= opts.min_confidence);
    if let Some(filter) = &opts.include_kinds {
        results.retain(|root| root.kinds.iter().any(|kind| filter.contains(kind)));
    }
    results.sort_by(|left, right| left.path.cmp(&right.path));
    Ok(results)
}

pub fn summarize_roots(roots: &[DiscoveredRoot]) -> DiscoverSummary {
    let mut by_kind: BTreeMap<RootKind, usize> = BTreeMap::new();
    let mut by_confidence: BTreeMap<Confidence, usize> = BTreeMap::new();
    for root in roots {
        *by_confidence.entry(root.confidence).or_insert(0) += 1;
        for kind in &root.kinds {
            *by_kind.entry(*kind).or_insert(0) += 1;
        }
    }
    DiscoverSummary {
        total: roots.len(),
        by_kind,
        by_confidence,
    }
}

#[derive(Clone, Debug, Eq, PartialEq)]
pub struct DiscoverSummary {
    pub total: usize,
    pub by_kind: BTreeMap<RootKind, usize>,
    pub by_confidence: BTreeMap<Confidence, usize>,
}

pub fn render_discover_markdown(
    root: &Path,
    roots: &[DiscoveredRoot],
    summary: &DiscoverSummary,
) -> String {
    let mut out = String::new();
    out.push_str("# Projd Discover Report\n\n");
    out.push_str(&format!("- Root: `{}`\n", root.display()));
    out.push_str(&format!("- Total project roots: {}\n", summary.total));
    if !summary.by_confidence.is_empty() {
        let parts: Vec<String> = [Confidence::Strong, Confidence::Medium, Confidence::Weak]
            .iter()
            .filter_map(|level| {
                summary
                    .by_confidence
                    .get(level)
                    .map(|count| format!("{}: {}", level.token(), count))
            })
            .collect();
        if !parts.is_empty() {
            out.push_str(&format!("- Confidence: {}\n", parts.join(" / ")));
        }
    }
    if !summary.by_kind.is_empty() {
        out.push_str("- Kinds:\n");
        for (kind, count) in &summary.by_kind {
            out.push_str(&format!("  - {}: {}\n", kind.token(), count));
        }
    }
    out.push('\n');

    if roots.is_empty() {
        out.push_str("No project roots found.\n");
        return out;
    }

    out.push_str("| Path | Kinds | Confidence | Category |\n");
    out.push_str("| --- | --- | --- | --- |\n");
    for entry in roots {
        let rel = relative_display(root, &entry.path);
        let kinds: Vec<&'static str> = entry.kinds.iter().map(|k| k.token()).collect();
        let mut cats: BTreeSet<&'static str> = BTreeSet::new();
        for kind in &entry.kinds {
            cats.insert(category_token(kind.category()));
        }
        out.push_str(&format!(
            "| `{}` | {} | {} | {} |\n",
            rel,
            kinds.join(", "),
            entry.confidence.token(),
            cats.into_iter().collect::<Vec<_>>().join(", "),
        ));
    }
    out
}

pub fn render_discover_json(
    root: &Path,
    roots: &[DiscoveredRoot],
    summary: &DiscoverSummary,
) -> Result<String> {
    #[derive(Serialize)]
    struct ReportView<'a> {
        root: &'a Path,
        total: usize,
        by_kind: BTreeMap<&'static str, usize>,
        by_confidence: BTreeMap<&'static str, usize>,
        roots: Vec<RootView<'a>>,
    }

    #[derive(Serialize)]
    struct RootView<'a> {
        path: &'a Path,
        relative_path: String,
        kinds: Vec<&'static str>,
        confidence: &'static str,
        category: Vec<&'static str>,
    }

    let report = ReportView {
        root,
        total: summary.total,
        by_kind: summary
            .by_kind
            .iter()
            .map(|(kind, count)| (kind.token(), *count))
            .collect(),
        by_confidence: summary
            .by_confidence
            .iter()
            .map(|(level, count)| (level.token(), *count))
            .collect(),
        roots: roots
            .iter()
            .map(|entry| {
                let mut cats: BTreeSet<&'static str> = BTreeSet::new();
                for kind in &entry.kinds {
                    cats.insert(category_token(kind.category()));
                }
                RootView {
                    path: &entry.path,
                    relative_path: relative_display(root, &entry.path),
                    kinds: entry.kinds.iter().map(|kind| kind.token()).collect(),
                    confidence: entry.confidence.token(),
                    category: cats.into_iter().collect(),
                }
            })
            .collect(),
    };

    serde_json::to_string_pretty(&report).context("failed to serialize discover report as JSON")
}

pub fn category_token(category: RootCategory) -> &'static str {
    match category {
        RootCategory::Vcs => "vcs",
        RootCategory::Workspace => "workspace",
        RootCategory::Package => "package",
        RootCategory::Docs => "docs",
        RootCategory::Data => "data",
    }
}

pub fn relative_display(base: &Path, target: &Path) -> String {
    match target.strip_prefix(base) {
        Ok(rel) => {
            let s = rel.display().to_string();
            if s.is_empty() { ".".to_string() } else { s }
        }
        Err(_) => target.display().to_string(),
    }
}

enum DirClass {
    Root {
        kinds: Vec<RootKind>,
        confidence: Confidence,
    },
    Container,
    Skip,
}

fn push_children(dir: &Path, depth: usize, stack: &mut Vec<(PathBuf, usize)>) {
    let Ok(entries) = fs::read_dir(dir) else {
        return;
    };
    let mut children: Vec<PathBuf> = Vec::new();
    for entry in entries.flatten() {
        let Ok(file_type) = entry.file_type() else {
            continue;
        };
        if file_type.is_symlink() || !file_type.is_dir() {
            continue;
        }
        if should_skip_directory(&entry.file_name()) {
            continue;
        }
        children.push(entry.path());
    }
    children.sort();
    for path in children.into_iter().rev() {
        stack.push((path, depth + 1));
    }
}

fn classify_dir(dir: &Path) -> DirClass {
    if !dir.is_dir() {
        return DirClass::Skip;
    }
    if dir.file_name().map(should_skip_directory).unwrap_or(false) {
        return DirClass::Skip;
    }

    let mut kinds: Vec<RootKind> = Vec::new();
    let mut confidence = Confidence::Weak;

    let record =
        |kind: RootKind, level: Confidence, kinds: &mut Vec<RootKind>, conf: &mut Confidence| {
            kinds.push(kind);
            if level > *conf {
                *conf = level;
            }
        };

    // VCS roots.
    if dir.join(".git").exists() {
        record(
            RootKind::GitRepo,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
    }
    if dir.join(".hg").is_dir() {
        record(
            RootKind::HgRepo,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
    }
    if dir.join(".svn").is_dir() {
        record(
            RootKind::SvnRepo,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
    }
    if dir.join(".fslckout").is_file() || dir.join("_FOSSIL_").is_file() {
        record(
            RootKind::FossilRepo,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
    }
    if dir.join(".bzr").is_dir() {
        record(
            RootKind::BzrRepo,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
    }

    // Cargo.
    let cargo_toml = dir.join("Cargo.toml");
    if cargo_toml.is_file() {
        let value = read_toml(&cargo_toml);
        let has_workspace = value.as_ref().and_then(|v| v.get("workspace")).is_some();
        let has_package = value.as_ref().and_then(|v| v.get("package")).is_some();
        if has_workspace {
            record(
                RootKind::CargoWorkspace,
                Confidence::Strong,
                &mut kinds,
                &mut confidence,
            );
        }
        if has_package {
            record(
                RootKind::CargoPackage,
                Confidence::Strong,
                &mut kinds,
                &mut confidence,
            );
        }
        if !has_workspace && !has_package {
            record(
                RootKind::CargoPackage,
                Confidence::Medium,
                &mut kinds,
                &mut confidence,
            );
        }
    }

    // Node / JavaScript ecosystems.
    let package_json = dir.join("package.json");
    if package_json.is_file() {
        record(
            RootKind::NpmPackage,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
        if let Some(value) = read_json(&package_json) {
            if value.get("workspaces").is_some() {
                record(
                    RootKind::YarnWorkspace,
                    Confidence::Strong,
                    &mut kinds,
                    &mut confidence,
                );
            }
        }
    }
    if dir.join("pnpm-workspace.yaml").is_file() || dir.join("pnpm-workspace.yml").is_file() {
        record(
            RootKind::PnpmWorkspace,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
    }
    if dir.join("lerna.json").is_file() {
        record(
            RootKind::LernaWorkspace,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
    }
    if dir.join("turbo.json").is_file() {
        record(
            RootKind::TurboWorkspace,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
    }

    // Python.
    if dir.join("pyproject.toml").is_file() {
        record(
            RootKind::PythonProject,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
    }

    // Go.
    if dir.join("go.work").is_file() {
        record(
            RootKind::GoWorkspace,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
    }
    if dir.join("go.mod").is_file() {
        record(
            RootKind::GoModule,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
    }

    // JVM.
    if dir.join("pom.xml").is_file() {
        record(
            RootKind::MavenProject,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
    }
    let gradle_settings =
        dir.join("settings.gradle").is_file() || dir.join("settings.gradle.kts").is_file();
    let gradle_build = dir.join("build.gradle").is_file() || dir.join("build.gradle.kts").is_file();
    if gradle_settings {
        record(
            RootKind::GradleWorkspace,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
    } else if gradle_build {
        record(
            RootKind::GradleProject,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
    }

    // C / C++.
    if dir.join("CMakeLists.txt").is_file() {
        record(
            RootKind::CMakeProject,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
    }

    // Ruby / PHP / Elixir.
    if dir.join("Gemfile").is_file() {
        record(
            RootKind::RubyProject,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
    }
    if dir.join("composer.json").is_file() {
        record(
            RootKind::PhpProject,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
    }
    if dir.join("mix.exs").is_file() {
        record(
            RootKind::ElixirProject,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
    }

    if dir_has_extension(dir, &["csproj", "fsproj", "vbproj", "sln"]) {
        record(
            RootKind::DotnetProject,
            Confidence::Strong,
            &mut kinds,
            &mut confidence,
        );
    }

    // Documentation / static site frameworks.
    if dir.join("book.toml").is_file() {
        record(
            RootKind::MdBook,
            Confidence::Medium,
            &mut kinds,
            &mut confidence,
        );
    }
    if dir.join("mkdocs.yml").is_file() || dir.join("mkdocs.yaml").is_file() {
        record(
            RootKind::MkDocs,
            Confidence::Medium,
            &mut kinds,
            &mut confidence,
        );
    }
    if dir.join("_config.yml").is_file() {
        record(
            RootKind::Jekyll,
            Confidence::Medium,
            &mut kinds,
            &mut confidence,
        );
    }
    if dir.join("conf.py").is_file() && dir.join("index.rst").is_file() {
        record(
            RootKind::Sphinx,
            Confidence::Medium,
            &mut kinds,
            &mut confidence,
        );
    }
    if file_with_any_extension(dir, "docusaurus.config", &["js", "ts", "mjs", "cjs"]) {
        record(
            RootKind::Docusaurus,
            Confidence::Medium,
            &mut kinds,
            &mut confidence,
        );
    }
    if dir.join("hugo.toml").is_file()
        || dir.join("hugo.yaml").is_file()
        || dir.join("hugo.json").is_file()
    {
        record(
            RootKind::Hugo,
            Confidence::Medium,
            &mut kinds,
            &mut confidence,
        );
    }
    if file_with_any_extension(dir, "gatsby-config", &["js", "ts", "mjs", "cjs"]) {
        record(
            RootKind::Gatsby,
            Confidence::Medium,
            &mut kinds,
            &mut confidence,
        );
    }
    if file_with_any_extension(dir, "astro.config", &["js", "ts", "mjs", "cjs"]) {
        record(
            RootKind::Astro,
            Confidence::Medium,
            &mut kinds,
            &mut confidence,
        );
    }
    if dir.join("docfx.json").is_file() {
        record(
            RootKind::DocFx,
            Confidence::Medium,
            &mut kinds,
            &mut confidence,
        );
    }

    // Data.
    if dir.join("dvc.yaml").is_file() || dir.join(".dvc").is_dir() {
        record(
            RootKind::DvcDataset,
            Confidence::Medium,
            &mut kinds,
            &mut confidence,
        );
    }

    kinds.sort();
    kinds.dedup();

    if kinds.is_empty() {
        DirClass::Container
    } else {
        DirClass::Root { kinds, confidence }
    }
}

fn read_toml(path: &Path) -> Option<toml::Value> {
    let content = fs::read_to_string(path).ok()?;
    toml::from_str(&content).ok()
}

fn read_json(path: &Path) -> Option<serde_json::Value> {
    serde_json::from_str(&fs::read_to_string(path).ok()?).ok()
}

fn dir_has_extension(dir: &Path, exts: &[&str]) -> bool {
    let Ok(entries) = fs::read_dir(dir) else {
        return false;
    };
    for entry in entries.flatten() {
        let path = entry.path();
        if let Some(ext) = path.extension().and_then(|s| s.to_str()) {
            let lower = ext.to_ascii_lowercase();
            if exts.iter().any(|allowed| *allowed == lower) {
                return true;
            }
        }
    }
    false
}

fn file_with_any_extension(dir: &Path, stem: &str, exts: &[&str]) -> bool {
    exts.iter()
        .any(|ext| dir.join(format!("{stem}.{ext}")).is_file())
}

fn should_skip_directory(name: &OsStr) -> bool {
    let Some(s) = name.to_str() else {
        return false;
    };
    matches!(
        s,
        ".git"
            | ".hg"
            | ".svn"
            | ".bzr"
            | "target"
            | "node_modules"
            | ".venv"
            | "venv"
            | "dist"
            | "build"
            | "out"
            | "vendor"
            | ".idea"
            | ".vscode"
            | ".cache"
            | "__pycache__"
            | ".gradle"
            | ".tox"
            | ".pytest_cache"
            | ".mypy_cache"
            | ".next"
            | ".nuxt"
            | ".turbo"
            | ".parcel-cache"
            | ".docusaurus"
            | "_site"
            | ".jekyll-cache"
            | "Pods"
            | "DerivedData"
    )
}

#[cfg(test)]
mod tests {
    use super::*;
    use std::fs;
    use tempfile::tempdir;

    fn write_file(dir: &Path, name: &str, content: &str) {
        let path = dir.join(name);
        if let Some(parent) = path.parent() {
            fs::create_dir_all(parent).unwrap();
        }
        fs::write(path, content).unwrap();
    }

    fn make_dir(dir: &Path, name: &str) {
        fs::create_dir_all(dir.join(name)).unwrap();
    }

    #[test]
    fn empty_dir_yields_no_roots() {
        let tmp = tempdir().unwrap();
        let result = discover_roots(tmp.path(), &DiscoverOptions::default()).unwrap();
        assert!(result.is_empty());
    }

    #[test]
    fn single_cargo_package() {
        let tmp = tempdir().unwrap();
        write_file(
            tmp.path(),
            "Cargo.toml",
            "[package]\nname = \"x\"\nversion = \"0.1.0\"\n",
        );
        let result = discover_roots(tmp.path(), &DiscoverOptions::default()).unwrap();
        assert_eq!(result.len(), 1);
        assert!(result[0].kinds.contains(&RootKind::CargoPackage));
        assert_eq!(result[0].confidence, Confidence::Strong);
    }

    #[test]
    fn workspace_default_does_not_expand() {
        let tmp = tempdir().unwrap();
        let root = tmp.path();
        write_file(
            root,
            "Cargo.toml",
            "[workspace]\nmembers = [\"a\", \"b\"]\n",
        );
        for member in ["a", "b"] {
            write_file(
                &root.join(member),
                "Cargo.toml",
                "[package]\nname = \"m\"\nversion = \"0.1.0\"\n",
            );
        }
        let result = discover_roots(root, &DiscoverOptions::default()).unwrap();
        assert_eq!(result.len(), 1);
        assert!(result[0].kinds.contains(&RootKind::CargoWorkspace));
    }

    #[test]
    fn workspace_expand_yields_members() {
        let tmp = tempdir().unwrap();
        let root = tmp.path();
        write_file(
            root,
            "Cargo.toml",
            "[workspace]\nmembers = [\"a\", \"b\"]\n",
        );
        for member in ["a", "b"] {
            write_file(
                &root.join(member),
                "Cargo.toml",
                "[package]\nname = \"m\"\nversion = \"0.1.0\"\n",
            );
        }
        let opts = DiscoverOptions {
            expand_workspaces: true,
            ..DiscoverOptions::default()
        };
        let result = discover_roots(root, &opts).unwrap();
        assert_eq!(result.len(), 3);
    }

    #[test]
    fn side_by_side_repos_are_separate_roots() {
        let tmp = tempdir().unwrap();
        let root = tmp.path();
        for repo in ["repoA", "repoB"] {
            let dir = root.join(repo);
            fs::create_dir_all(&dir).unwrap();
            fs::create_dir(dir.join(".git")).unwrap();
        }
        let result = discover_roots(root, &DiscoverOptions::default()).unwrap();
        assert_eq!(result.len(), 2);
        assert!(result.iter().all(|r| r.kinds.contains(&RootKind::GitRepo)));
    }

    #[test]
    fn mdbook_default_passes_medium_filter() {
        let tmp = tempdir().unwrap();
        write_file(tmp.path(), "book.toml", "[book]\ntitle = \"x\"\n");
        let result = discover_roots(tmp.path(), &DiscoverOptions::default()).unwrap();
        assert_eq!(result.len(), 1);
        assert!(result[0].kinds.contains(&RootKind::MdBook));
        assert_eq!(result[0].confidence, Confidence::Medium);
    }

    #[test]
    fn min_confidence_strong_filters_mdbook() {
        let tmp = tempdir().unwrap();
        write_file(tmp.path(), "book.toml", "[book]\ntitle = \"x\"\n");
        let opts = DiscoverOptions {
            min_confidence: Confidence::Strong,
            ..DiscoverOptions::default()
        };
        let result = discover_roots(tmp.path(), &opts).unwrap();
        assert!(result.is_empty());
    }

    #[test]
    fn nested_vcs_default_skipped() {
        // Use a non-skipped directory name so we test the nested_vcs gate, not
        // the directory-skip list (which intentionally excludes `vendor/`).
        let tmp = tempdir().unwrap();
        let outer = tmp.path();
        fs::create_dir(outer.join(".git")).unwrap();
        make_dir(outer, "third_party/dep");
        fs::create_dir(outer.join("third_party/dep/.git")).unwrap();

        let result = discover_roots(outer, &DiscoverOptions::default()).unwrap();
        assert_eq!(result.len(), 1);

        let opts = DiscoverOptions {
            nested_vcs: true,
            ..DiscoverOptions::default()
        };
        let result = discover_roots(outer, &opts).unwrap();
        assert_eq!(result.len(), 2);
    }

    #[test]
    fn include_kind_filter() {
        let tmp = tempdir().unwrap();
        let root = tmp.path();
        let a = root.join("a");
        let b = root.join("b");
        fs::create_dir_all(&a).unwrap();
        fs::create_dir_all(&b).unwrap();
        write_file(
            &a,
            "Cargo.toml",
            "[package]\nname=\"a\"\nversion=\"0.1.0\"\n",
        );
        write_file(&b, "package.json", "{\"name\":\"b\"}\n");

        let mut filter = BTreeSet::new();
        filter.insert(RootKind::NpmPackage);
        let opts = DiscoverOptions {
            include_kinds: Some(filter),
            ..DiscoverOptions::default()
        };
        let result = discover_roots(root, &opts).unwrap();
        assert_eq!(result.len(), 1);
        assert!(result[0].kinds.contains(&RootKind::NpmPackage));
    }

    #[test]
    fn skips_node_modules_and_target() {
        let tmp = tempdir().unwrap();
        let root = tmp.path();
        // hidden a fake project inside node_modules; must not be reported
        let buried = root.join("node_modules").join("inner");
        fs::create_dir_all(&buried).unwrap();
        write_file(&buried, "package.json", "{\"name\":\"inner\"}\n");

        let result = discover_roots(root, &DiscoverOptions::default()).unwrap();
        assert!(result.is_empty());
    }

    #[test]
    fn from_token_round_trip() {
        for kind in [
            RootKind::GitRepo,
            RootKind::CargoPackage,
            RootKind::MdBook,
            RootKind::DvcDataset,
        ] {
            assert_eq!(RootKind::from_token(kind.token()), Some(kind));
        }
        assert_eq!(RootKind::from_token("git"), Some(RootKind::GitRepo));
        assert_eq!(RootKind::from_token("cargo"), Some(RootKind::CargoPackage));
        assert_eq!(RootKind::from_token("nope"), None);
    }
}