fallow_config/workspace/

mod.rs

mod deno_json;
mod diagnostics;
mod npm_overrides;
mod package_json;
mod parsers;
mod pnpm_catalog;
mod pnpm_overrides;
mod tsconfig;

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

use rustc_hash::FxHashMap;
use schemars::JsonSchema;
use serde::{Deserialize, Serialize};

pub use deno_json::{
    DirManifestProbe, dir_has_deno_json, dir_has_package_manifest, is_deno_without_node_modules,
    load_deno_import_map, load_dir_package_json, load_member_package_manifest,
    load_root_deno_workspace_patterns, probe_dir_manifest,
};
#[cfg(test)]
pub use diagnostics::capture_workspace_warnings;
pub use diagnostics::{
    WorkspaceDiagnostic, WorkspaceDiagnosticKind, WorkspaceLoadError, append_workspace_diagnostics,
    clear_analysis_stage_diagnostics, clear_health_stage_diagnostics,
    health_stage_workspace_diagnostics, missing_node_modules_diagnostic, node_modules_missing,
    record_plugin_config_diagnostics, record_source_parse_degradations,
    record_source_read_failures, record_workspace_diagnostics, registry_diagnostics_to_fold,
    replace_source_discovery_diagnostics, stash_workspace_diagnostics, workspace_diagnostics_for,
};
use diagnostics::{emit_diagnostics, is_ignored_workspace_dir, is_skip_listed_dir};
pub use npm_overrides::{parse_bun_package_json_resolutions, parse_npm_package_json_overrides};
pub use package_json::{NapiConfig, PackageJson};
pub use parsers::parse_tsconfig_root_dir;
use parsers::{
    expand_workspace_glob_with_diagnostics, parse_pnpm_workspace_yaml,
    parse_tsconfig_references_with_diagnostics,
};
pub use pnpm_catalog::{
    PnpmCatalog, PnpmCatalogData, PnpmCatalogEntry, PnpmCatalogGroup,
    parse_package_json_catalog_data, parse_pnpm_catalog_data,
};
pub use pnpm_overrides::{
    MisconfigReason, OverrideSource, ParsedOverrideKey, PnpmOverrideData, PnpmOverrideEntry,
    is_valid_override_value, override_misconfig_reason, override_source_label, parse_override_key,
    parse_pnpm_package_json_overrides, parse_pnpm_workspace_overrides,
};
pub use tsconfig::{TsconfigOutputMap, TsconfigOutputResolution};

/// Workspace configuration for monorepo support.
#[derive(Debug, Clone, Deserialize, Serialize, JsonSchema)]
pub struct WorkspaceConfig {
    /// Additional workspace patterns (beyond what's in root package.json).
    ///
    /// `packages` is accepted as a back-compat alias: an older `fallow init --toml`
    /// wrote `[workspaces]` with a `packages` key, and reading it as `patterns`
    /// keeps those existing configs scoping correctly. schemars omits serde
    /// aliases, so `schema.json` documents only `patterns`.
    #[serde(default, alias = "packages")]
    pub patterns: Vec<String>,
    /// Git baseline refs keyed by workspace roots, written relative to the project root as `fallow list --workspaces` prints them. Scopes `check`, `dead-code`, `dupes` and editor diagnostics; `health`, `security` and `audit` ignore it. A global changed-since request takes precedence. Unlisted workspaces and root files remain in full scope. A key that names no workspace, or a ref that Git cannot resolve, leaves every package in full scope with a warning.
    #[serde(
        default,
        rename = "changedSince",
        skip_serializing_if = "BTreeMap::is_empty"
    )]
    pub changed_since: BTreeMap<String, String>,
}

/// Discovered workspace info from package.json, deno.json, pnpm-workspace.yaml,
/// or tsconfig.json references.
#[derive(Debug, Clone)]
pub struct WorkspaceInfo {
    /// Workspace root path.
    pub root: PathBuf,
    /// Package name from package.json or deno.json.
    pub name: String,
    /// Whether this workspace is depended on by other workspaces.
    pub is_internal_dependency: bool,
}

/// Return whether a workspace package name is selected by `publicPackages`.
///
/// Exact names and glob patterns share one config-owned matcher so analysis
/// surfaces cannot disagree about which workspace APIs are public. Invalid
/// patterns do not match, preserving the existing fail-closed behavior.
#[must_use]
pub fn workspace_is_public(name: &str, public_packages: &[String]) -> bool {
    public_packages.iter().any(|pattern| {
        name == pattern
            || globset::Glob::new(pattern).is_ok_and(|glob| glob.compile_matcher().is_match(name))
    })
}

/// Discover all workspace packages in a monorepo.
///
/// Sources (additive, deduplicated by canonical path):
/// 1. `package.json` `workspaces` field
/// 2. `pnpm-workspace.yaml` `packages` field
/// 3. `deno.json` / `deno.jsonc` `workspace` field
/// 4. `tsconfig.json` `references` field (TypeScript project references)
/// 5. `link:` and `file:` dependency targets in the root `package.json`
///    (yarn-era monorepos without a `workspaces` field)
///
/// Back-compat wrapper: drops any diagnostics and silently treats a malformed
/// root `package.json` as "no workspaces". New callers should use
/// [`discover_workspaces_with_diagnostics`] to receive typed
/// [`WorkspaceDiagnostic`] values and to surface root-malformed errors as
/// hard exits.
///
/// This wrapper goes through the silent collector path that does NOT call
/// `emit_diagnostics` (private helper in `crates/config/src/workspace/diagnostics.rs`
/// that does the `tracing::warn!` emission). Without that split, sibling
/// callers in `core/src/lib.rs` (analyze) and `core/src/discover/mod.rs`
/// (file discovery) would re-emit `tracing::warn!` on paths the user already
/// excluded via `ignorePatterns`, because the back-compat wrapper has no
/// access to the user's globset.
#[must_use]
pub fn discover_workspaces(root: &Path) -> Vec<WorkspaceInfo> {
    collect_workspaces_and_diagnostics(root, &crate::IgnorePatternSet::empty())
        .map(|(workspaces, _)| workspaces)
        .unwrap_or_default()
}

/// Discover workspace packages and return any diagnostics produced along the
/// way.
///
/// Replaces the four silent-drop sites in [`discover_workspaces`] with typed
/// [`WorkspaceDiagnostic`] values:
/// - malformed declared-workspace `package.json` (warn-and-continue),
/// - glob match resolving to a directory without `package.json` (warn,
///   filtered through `ignore_patterns` and an extended skip list),
/// - malformed `tsconfig.json` (warn-and-continue),
/// - tsconfig `references[].path` pointing to a missing directory (warn).
///
/// `ignore_patterns` mirrors the precedent in
/// [`find_undeclared_workspaces_with_ignores`]: directories the user already
/// excluded do not trigger a redundant diagnostic.
///
/// Returns [`WorkspaceLoadError::MalformedRootPackageJson`] when the root
/// `package.json` exists but fails to parse: without a parseable root, no
/// workspace patterns can be collected and the analysis output would be
/// fiction. The CLI surfaces this as exit 2.
///
/// Shallow-scan fallback candidates (`collect_shallow_workspace_candidate`)
/// stay silent: the user did not declare them, so a stray malformed
/// `package.json` two levels deep in a `tools/scratch/` directory should not
/// produce noise.
///
/// # Errors
///
/// Returns [`WorkspaceLoadError`] when the project root's `package.json`
/// exists but is not valid JSON. Callers map this to a hard exit.
pub fn discover_workspaces_with_diagnostics(
    root: &Path,
    ignore_patterns: &crate::IgnorePatternSet,
) -> Result<(Vec<WorkspaceInfo>, Vec<WorkspaceDiagnostic>), WorkspaceLoadError> {
    let (workspaces, diagnostics) = collect_workspaces_and_diagnostics(root, ignore_patterns)?;

    emit_diagnostics(root, &diagnostics);

    Ok((workspaces, diagnostics))
}

/// Collect workspaces and diagnostics without emitting `tracing::warn!`.
///
/// Both [`discover_workspaces_with_diagnostics`] (which adds the emit step)
/// and [`discover_workspaces`] (which drops both diagnostics and emission)
/// route through this function. Keeping emission in the public top-level
/// only means downstream callers that have no access to the user's
/// `ignorePatterns` cannot accidentally re-emit warnings on paths the user
/// already excluded.
///
/// The returned diagnostics are deduplicated on the whole `(kind, path)`, the
/// way every downstream fold is. The sources are additive, so a repository that
/// declares one glob in both `package.json` `workspaces` and
/// `pnpm-workspace.yaml` `packages` walks it twice and reports every
/// package-less directory under it twice. Folding here rather than only in the
/// process registry is what keeps the returned list, the `workspaces` and
/// `list --workspaces` envelopes, the MCP `project_info` tool and the
/// aggregated stderr warning (whose per-pattern grouping counts the entries it
/// is handed) agreeing on one entry per distinct matching pattern (issue
/// #2366). Two overlapping globs (`["packages/*", "packages/*/*"]`) still
/// report the same directory once per `pattern`: the payload is part of the
/// key.
fn collect_workspaces_and_diagnostics(
    root: &Path,
    ignore_patterns: &crate::IgnorePatternSet,
) -> Result<(Vec<WorkspaceInfo>, Vec<WorkspaceDiagnostic>), WorkspaceLoadError> {
    let mut diagnostics = Vec::new();
    let patterns = collect_workspace_patterns(root)?;
    let canonical_root = dunce::canonicalize(root).unwrap_or_else(|_| root.to_path_buf());
    let mut manifest_cache = ManifestCache::default();

    let mut workspaces = collect_declared_workspaces(
        root,
        &patterns,
        &canonical_root,
        ignore_patterns,
        &mut diagnostics,
        &mut manifest_cache,
    );
    workspaces.extend(collect_link_dependency_workspaces(
        root,
        &canonical_root,
        ignore_patterns,
        &mut diagnostics,
        &mut manifest_cache,
    ));
    if patterns.is_empty() {
        workspaces.extend(collect_shallow_package_workspaces(
            root,
            &canonical_root,
            &mut manifest_cache,
        ));
    }

    if !workspaces.is_empty() {
        mark_internal_dependencies(&mut workspaces);
    }
    let workspaces = workspaces.into_iter().map(|(ws, _)| ws).collect();
    Ok((
        workspaces,
        fallow_types::workspace::dedupe_workspace_diagnostics(diagnostics),
    ))
}

/// Workspaces that a workspace glob (`package.json`, `pnpm-workspace.yaml`,
/// Deno) or a `tsconfig.json` reference declares.
fn collect_declared_workspaces(
    root: &Path,
    patterns: &[String],
    canonical_root: &Path,
    ignore_patterns: &crate::IgnorePatternSet,
    diagnostics: &mut Vec<WorkspaceDiagnostic>,
    manifest_cache: &mut ManifestCache,
) -> Vec<(WorkspaceInfo, Vec<String>)> {
    let mut workspaces = expand_patterns_to_workspaces(
        root,
        patterns,
        canonical_root,
        ignore_patterns,
        diagnostics,
        manifest_cache,
    );
    workspaces.extend(collect_tsconfig_workspaces(
        root,
        canonical_root,
        ignore_patterns,
        diagnostics,
        manifest_cache,
    ));
    workspaces
}

/// Names of root `link:` and `file:` dependencies whose spec is the only
/// declaration of a discovered workspace.
///
/// In a yarn-era monorepo without a `workspaces` field, the root entry is how
/// the repository declares the package. Removing it removes the package from
/// the analysis, so the unused-dependency pass must not report it. When a
/// workspace glob, `pnpm-workspace.yaml`, a Deno workspace or a `tsconfig.json`
/// reference also declares the target, the entry is an ordinary dependency.
/// A malformed root manifest gives an empty set, because discovery reports it.
#[must_use]
pub fn link_only_workspace_dependencies(
    root: &Path,
    ignore_patterns: &crate::IgnorePatternSet,
    workspaces: &[WorkspaceInfo],
) -> rustc_hash::FxHashSet<String> {
    let Ok(root_pkg) = PackageJson::load(&root.join("package.json")) else {
        return rustc_hash::FxHashSet::default();
    };
    let links = root_pkg.local_link_dependencies();
    if links.is_empty() || workspaces.is_empty() {
        return rustc_hash::FxHashSet::default();
    }
    let canonical = |path: &Path| dunce::canonicalize(path).unwrap_or_else(|_| path.to_path_buf());
    let workspace_roots: rustc_hash::FxHashSet<PathBuf> =
        workspaces.iter().map(|ws| canonical(&ws.root)).collect();
    let linked: Vec<(String, PathBuf)> = links
        .into_iter()
        .map(|(name, target)| (name, canonical(&root.join(target))))
        .filter(|(_, dir)| workspace_roots.contains(dir))
        .collect();
    if linked.is_empty() {
        return rustc_hash::FxHashSet::default();
    }

    let patterns = collect_workspace_patterns(root).unwrap_or_default();
    let declared: rustc_hash::FxHashSet<PathBuf> = collect_declared_workspaces(
        root,
        &patterns,
        &canonical(root),
        ignore_patterns,
        &mut Vec::new(),
        &mut ManifestCache::default(),
    )
    .into_iter()
    .map(|(ws, _)| canonical(&ws.root))
    .collect();
    linked
        .into_iter()
        .filter(|(_, dir)| !declared.contains(dir))
        .map(|(name, _)| name)
        .collect()
}

/// Find directories containing `package.json` that are not declared as workspaces.
///
/// Only meaningful in monorepos that declare workspaces (via `package.json` `workspaces`
/// field or `pnpm-workspace.yaml`). Scans up to two directory levels deep, skipping
/// hidden directories, `node_modules`, and `build`.
#[must_use]
pub fn find_undeclared_workspaces(
    root: &Path,
    declared: &[WorkspaceInfo],
) -> Vec<WorkspaceDiagnostic> {
    find_undeclared_workspaces_with_ignores(root, declared, &crate::IgnorePatternSet::empty())
}

/// Find directories containing `package.json` that are not declared as workspaces,
/// excluding candidates covered by the supplied ignore globset.
///
/// This is the ignore-aware variant used by the full analyzer after config
/// resolution. See [`find_undeclared_workspaces`] for the compatibility wrapper.
///
/// Directories whose project-root-relative path matches `ignore_patterns` are skipped
/// so users who already excluded a path via `ignorePatterns` don't see a redundant
/// "not declared as workspace" warning. See issue #193.
#[must_use]
pub fn find_undeclared_workspaces_with_ignores(
    root: &Path,
    declared: &[WorkspaceInfo],
    ignore_patterns: &crate::IgnorePatternSet,
) -> Vec<WorkspaceDiagnostic> {
    let patterns = collect_workspace_patterns(root).unwrap_or_default();
    if patterns.is_empty() {
        return Vec::new();
    }

    let declared_roots: rustc_hash::FxHashSet<PathBuf> = declared
        .iter()
        .map(|w| dunce::canonicalize(&w.root).unwrap_or_else(|_| w.root.clone()))
        .collect();

    let canonical_root = dunce::canonicalize(root).unwrap_or_else(|_| root.to_path_buf());

    let mut undeclared = Vec::new();

    let Ok(top_entries) = std::fs::read_dir(root) else {
        return Vec::new();
    };

    for entry in top_entries.filter_map(Result::ok) {
        let path = entry.path();
        if !path.is_dir() || is_undeclared_scan_skip_dir(&entry.file_name().to_string_lossy()) {
            continue;
        }

        let mut input = UndeclaredScanInput {
            root,
            canonical_root: &canonical_root,
            declared_roots: &declared_roots,
            ignore_patterns,
            undeclared: &mut undeclared,
        };
        check_undeclared(
            &path,
            input.root,
            input.canonical_root,
            input.declared_roots,
            input.ignore_patterns,
            input.undeclared,
        );
        scan_child_dirs_for_undeclared(&path, &mut input);
    }

    undeclared
}

/// Whether an undeclared-workspace scan should skip a directory by leaf name.
fn is_undeclared_scan_skip_dir(name: &str) -> bool {
    name.starts_with('.') || name == "node_modules" || name == "build"
}

/// Borrowed inputs threaded through the second-level undeclared-workspace scan.
struct UndeclaredScanInput<'a> {
    root: &'a Path,
    canonical_root: &'a Path,
    declared_roots: &'a rustc_hash::FxHashSet<PathBuf>,
    ignore_patterns: &'a crate::IgnorePatternSet,
    undeclared: &'a mut Vec<WorkspaceDiagnostic>,
}

/// Check each immediate child of `parent` for an undeclared workspace.
fn scan_child_dirs_for_undeclared(parent: &Path, input: &mut UndeclaredScanInput<'_>) {
    let Ok(child_entries) = std::fs::read_dir(parent) else {
        return;
    };
    for child in child_entries.filter_map(Result::ok) {
        let child_path = child.path();
        if !child_path.is_dir() || is_undeclared_scan_skip_dir(&child.file_name().to_string_lossy())
        {
            continue;
        }
        check_undeclared(
            &child_path,
            input.root,
            input.canonical_root,
            input.declared_roots,
            input.ignore_patterns,
            input.undeclared,
        );
    }
}

/// Check a single directory for an undeclared workspace.
fn check_undeclared(
    dir: &Path,
    root: &Path,
    canonical_root: &Path,
    declared_roots: &rustc_hash::FxHashSet<PathBuf>,
    ignore_patterns: &crate::IgnorePatternSet,
    undeclared: &mut Vec<WorkspaceDiagnostic>,
) {
    if !dir_has_package_manifest(dir) {
        return;
    }
    let canonical = dunce::canonicalize(dir).unwrap_or_else(|_| dir.to_path_buf());
    if canonical == *canonical_root {
        return;
    }
    if declared_roots.contains(&canonical) {
        return;
    }
    let relative = dir.strip_prefix(root).unwrap_or(dir);
    let relative_str = relative.to_string_lossy().replace('\\', "/");
    if ignore_patterns.is_match(relative_str.as_str())
        || ignore_patterns.is_match(format!("{relative_str}/package.json").as_str())
        || ignore_patterns.is_match(format!("{relative_str}/deno.json").as_str())
        || ignore_patterns.is_match(format!("{relative_str}/deno.jsonc").as_str())
    {
        return;
    }
    undeclared.push(WorkspaceDiagnostic::new(
        root,
        dir.to_path_buf(),
        WorkspaceDiagnosticKind::UndeclaredWorkspace,
    ));
}

/// Collect glob patterns from `package.json` `workspaces`, `pnpm-workspace.yaml`,
/// and Deno `deno.json` / `deno.jsonc` `workspace`.
fn collect_workspace_patterns(root: &Path) -> Result<Vec<String>, WorkspaceLoadError> {
    let mut patterns = Vec::new();

    let pkg_path = root.join("package.json");
    if pkg_path.exists() {
        match PackageJson::load(&pkg_path) {
            Ok(pkg) => patterns.extend(pkg.workspace_patterns()),
            Err(error) => {
                return Err(WorkspaceLoadError::MalformedRootPackageJson {
                    path: pkg_path,
                    error,
                });
            }
        }
    }

    let pnpm_workspace = root.join("pnpm-workspace.yaml");
    if pnpm_workspace.exists()
        && let Ok(content) = std::fs::read_to_string(&pnpm_workspace)
    {
        patterns.extend(parse_pnpm_workspace_yaml(&content));
    }

    if let Some((_path, deno_patterns)) = load_root_deno_workspace_patterns(root)
        .map_err(|(path, error)| WorkspaceLoadError::MalformedRootDenoConfig { path, error })?
    {
        patterns.extend(deno_patterns);
    }

    Ok(patterns)
}

/// Memoized [`load_member_package_manifest`] outcomes for one discovery run.
///
/// The same directory is commonly matched by more than one workspace source
/// (identical globs in `package.json` and `pnpm-workspace.yaml`, tsconfig
/// references overlapping npm workspaces). Each manifest load costs a
/// `package.json` read plus Deno `deno.json` / `deno.jsonc` probes, so repeat
/// visits replay the memoized outcome instead of touching the filesystem again.
type ManifestCache = FxHashMap<PathBuf, Result<Option<(String, PackageJson, Vec<String>)>, String>>;

/// Load a member manifest through the per-discovery memo, preserving the
/// exact per-call outcome (including errors) on repeat visits.
fn load_member_package_manifest_cached(
    dir: &Path,
    cache: &mut ManifestCache,
) -> Result<Option<(String, PackageJson, Vec<String>)>, String> {
    if let Some(cached) = cache.get(dir) {
        return cached.clone();
    }
    let outcome = load_member_package_manifest(dir);
    cache.insert(dir.to_path_buf(), outcome.clone());
    outcome
}

/// Expand workspace glob patterns to discover workspace directories.
///
/// Handles positive/negated pattern splitting, glob matching, and package.json
/// loading for each matched directory.
fn expand_patterns_to_workspaces(
    root: &Path,
    patterns: &[String],
    canonical_root: &Path,
    ignore_patterns: &crate::IgnorePatternSet,
    diagnostics: &mut Vec<WorkspaceDiagnostic>,
    manifest_cache: &mut ManifestCache,
) -> Vec<(WorkspaceInfo, Vec<String>)> {
    if patterns.is_empty() {
        return Vec::new();
    }

    let mut workspaces = Vec::new();

    let (positive, negative): (Vec<&String>, Vec<&String>) =
        patterns.iter().partition(|p| !p.starts_with('!'));
    let negation_matchers: Vec<globset::GlobMatcher> = negative
        .iter()
        .filter_map(|p| {
            let stripped = p.strip_prefix('!').unwrap_or(p);
            globset::Glob::new(stripped)
                .ok()
                .map(|g| g.compile_matcher())
        })
        .collect();

    for pattern in &positive {
        let glob_pattern = if pattern.ends_with('/') {
            format!("{pattern}*")
        } else {
            (*pattern).clone()
        };

        let matched_dirs = expand_workspace_glob_with_diagnostics(
            root,
            pattern,
            &glob_pattern,
            canonical_root,
            ignore_patterns,
            diagnostics,
        );
        for (dir, canonical_dir) in matched_dirs {
            if canonical_dir == *canonical_root {
                continue;
            }
            if matches_negation(root, &dir, &negation_matchers) {
                continue;
            }
            register_matched_workspace(root, dir, &mut workspaces, diagnostics, manifest_cache);
        }
    }

    workspaces
}

/// Whether a matched directory is excluded by a negated workspace pattern.
fn matches_negation(root: &Path, dir: &Path, negation_matchers: &[globset::GlobMatcher]) -> bool {
    let relative = dir.strip_prefix(root).unwrap_or(dir);
    let relative_str = relative.to_string_lossy();
    negation_matchers
        .iter()
        .any(|m| m.is_match(relative_str.as_ref()))
}

/// Load a matched directory's package manifest (`package.json` or `deno.json`)
/// and push a workspace, or a malformed-package diagnostic on parse failure.
/// A manifest removed after glob expansion is treated as a stale match.
fn register_matched_workspace(
    root: &Path,
    dir: PathBuf,
    workspaces: &mut Vec<(WorkspaceInfo, Vec<String>)>,
    diagnostics: &mut Vec<WorkspaceDiagnostic>,
    manifest_cache: &mut ManifestCache,
) {
    match load_member_package_manifest_cached(&dir, manifest_cache) {
        Ok(Some((name, _pkg, dep_names))) => {
            workspaces.push((
                WorkspaceInfo {
                    root: dir,
                    name,
                    is_internal_dependency: false,
                },
                dep_names,
            ));
        }
        Ok(None) => {}
        Err(error) => {
            let diag = WorkspaceDiagnostic::new(
                root,
                dir,
                WorkspaceDiagnosticKind::MalformedPackageJson { error },
            );
            diagnostics.push(diag);
        }
    }
}

/// Discover workspaces from TypeScript project references in `tsconfig.json`.
///
/// Referenced directories are added as workspaces, supplementing npm/pnpm workspaces.
/// This enables cross-workspace resolution for TypeScript composite projects.
fn collect_tsconfig_workspaces(
    root: &Path,
    canonical_root: &Path,
    ignore_patterns: &crate::IgnorePatternSet,
    diagnostics: &mut Vec<WorkspaceDiagnostic>,
    manifest_cache: &mut ManifestCache,
) -> Vec<(WorkspaceInfo, Vec<String>)> {
    let mut workspaces = Vec::new();

    for dir in parse_tsconfig_references_with_diagnostics(root, ignore_patterns, diagnostics) {
        if let Some(workspace) =
            tsconfig_workspace_from_dir(root, dir, canonical_root, diagnostics, manifest_cache)
        {
            workspaces.push(workspace);
        }
    }

    workspaces
}

fn tsconfig_workspace_from_dir(
    root: &Path,
    dir: PathBuf,
    canonical_root: &Path,
    diagnostics: &mut Vec<WorkspaceDiagnostic>,
    manifest_cache: &mut ManifestCache,
) -> Option<(WorkspaceInfo, Vec<String>)> {
    let canonical_dir = dunce::canonicalize(&dir).unwrap_or_else(|_| dir.clone());
    if canonical_dir == *canonical_root || !canonical_dir.starts_with(canonical_root) {
        return None;
    }

    let (name, dep_names) =
        load_tsconfig_workspace_package(root, &dir, diagnostics, manifest_cache);
    Some((
        WorkspaceInfo {
            root: dir,
            name,
            is_internal_dependency: false,
        },
        dep_names,
    ))
}

fn load_tsconfig_workspace_package(
    root: &Path,
    dir: &Path,
    diagnostics: &mut Vec<WorkspaceDiagnostic>,
    manifest_cache: &mut ManifestCache,
) -> (String, Vec<String>) {
    match load_member_package_manifest_cached(dir, manifest_cache) {
        Ok(Some((name, _pkg, deps))) => (name, deps),
        Ok(None) => (dir_name(dir), Vec::new()),
        Err(error) => {
            let diag = WorkspaceDiagnostic::new(
                root,
                dir.to_path_buf(),
                WorkspaceDiagnosticKind::MalformedPackageJson { error },
            );
            diagnostics.push(diag);
            (dir_name(dir), Vec::new())
        }
    }
}

/// Discover workspaces from `link:` and `file:` dependency specs in the root
/// `package.json`.
///
/// Older yarn monorepos (for example Kibana before its pnpm move) declare no
/// `workspaces` field. The root lists each local package as
/// `"@scope/pkg": "link:path/to/pkg"`, often three to five directories deep,
/// out of reach of the shallow fallback scan. A target is a workspace only when
/// it is a directory inside the project root that holds a package manifest.
/// Targets outside the root, missing targets and tarballs are skipped without a
/// diagnostic, because a dependency spec is not a workspace declaration. A
/// target that source discovery does not walk is also skipped (see
/// [`is_skipped_link_target`] and [`link_targets_reached_by_source_walk`]). A
/// target with a malformed `package.json` gets the same diagnostic as a
/// declared workspace.
fn collect_link_dependency_workspaces(
    root: &Path,
    canonical_root: &Path,
    ignore_patterns: &crate::IgnorePatternSet,
    diagnostics: &mut Vec<WorkspaceDiagnostic>,
    manifest_cache: &mut ManifestCache,
) -> Vec<(WorkspaceInfo, Vec<String>)> {
    let Ok(root_pkg) = PackageJson::load(&root.join("package.json")) else {
        return Vec::new();
    };

    let candidates: Vec<PathBuf> = root_pkg
        .local_link_dependencies()
        .into_iter()
        .filter_map(|(_name, target)| dunce::canonicalize(root.join(target)).ok())
        .filter(|canonical_dir| {
            *canonical_dir != *canonical_root
                && canonical_dir.starts_with(canonical_root)
                && canonical_dir.is_dir()
                && !is_skipped_link_target(canonical_root, canonical_dir, ignore_patterns)
        })
        .collect();
    let walked = link_targets_reached_by_source_walk(canonical_root, &candidates);

    let mut workspaces = Vec::new();
    for canonical_dir in candidates {
        if !walked.contains(&canonical_dir) {
            continue;
        }
        let dir = normalize_link_target(root, canonical_root, &canonical_dir);
        register_matched_workspace(root, dir, &mut workspaces, diagnostics, manifest_cache);
    }
    workspaces
}

/// Whether a link target is under a path that source discovery skips by name.
///
/// A yalc copy (`file:.yalc/pkg`), a package under `node_modules`, a build
/// output or a path that `ignorePatterns` matches has no discovered source
/// files. A workspace there makes each import of the package an unresolved
/// import, so such a target stays an external package. The skip list is the
/// one the workspace glob expansion uses.
fn is_skipped_link_target(
    canonical_root: &Path,
    canonical_dir: &Path,
    ignore_patterns: &crate::IgnorePatternSet,
) -> bool {
    let Ok(relative) = canonical_dir.strip_prefix(canonical_root) else {
        return true;
    };
    relative
        .components()
        .any(|component| is_skip_listed_dir(&component.as_os_str().to_string_lossy()))
        || is_ignored_workspace_dir(relative, ignore_patterns)
}

/// The link targets that the source discovery walk reaches.
///
/// A target under a gitignored directory has no discovered source files, for
/// the same reason as a skip-listed target. The walk uses the ignore-file
/// settings of source discovery ([`crate::source_walk_builder`]), so nested
/// `.gitignore` files, `.git/info/exclude` and the global gitignore apply with
/// the same rules. The walk enters only the ancestor directories of the
/// targets, so it reads a small number of directories. A gitignore rule that
/// ignores files but not their directory (for example `*.ts`) does not stop
/// the walk, so such a target stays a workspace.
fn link_targets_reached_by_source_walk(
    canonical_root: &Path,
    targets: &[PathBuf],
) -> rustc_hash::FxHashSet<PathBuf> {
    let mut reached = rustc_hash::FxHashSet::default();
    if targets.is_empty() {
        return reached;
    }
    let mut on_path: rustc_hash::FxHashSet<PathBuf> = rustc_hash::FxHashSet::default();
    for target in targets {
        for ancestor in target.ancestors() {
            if ancestor == canonical_root || !on_path.insert(ancestor.to_path_buf()) {
                break;
            }
        }
    }
    let max_depth = targets
        .iter()
        .filter_map(|target| target.strip_prefix(canonical_root).ok())
        .map(|relative| relative.components().count())
        .max();
    let mut builder = crate::source_walk_builder(canonical_root);
    builder
        .max_depth(max_depth)
        .filter_entry(move |entry| entry.depth() == 0 || on_path.contains(entry.path()));
    let targets: rustc_hash::FxHashSet<&Path> = targets.iter().map(PathBuf::as_path).collect();
    for entry in builder.build().flatten() {
        if targets.contains(entry.path()) {
            reached.insert(entry.into_path());
        }
    }
    reached
}

/// Express a canonical link target under the caller's `root` spelling.
///
/// A spec such as `link:./pkg/../pkg` or a symlinked root would otherwise give
/// a workspace root that does not prefix-match the discovered source paths.
fn normalize_link_target(root: &Path, canonical_root: &Path, canonical_dir: &Path) -> PathBuf {
    canonical_dir
        .strip_prefix(canonical_root)
        .map_or_else(|_| canonical_dir.to_path_buf(), |rel| root.join(rel))
}

/// Discover shallow package workspaces when no explicit workspace config exists.
///
/// Scans direct children of the project root and their immediate children for
/// `package.json` files. This catches repos that contain multiple standalone
/// packages (for example `benchmarks/` or `editors/vscode/`) without declaring
/// npm/pnpm workspaces at the root.
fn collect_shallow_package_workspaces(
    root: &Path,
    canonical_root: &Path,
    manifest_cache: &mut ManifestCache,
) -> Vec<(WorkspaceInfo, Vec<String>)> {
    let mut workspaces = Vec::new();
    let Ok(top_entries) = std::fs::read_dir(root) else {
        return workspaces;
    };

    for entry in top_entries.filter_map(Result::ok) {
        let path = entry.path();
        if !path.is_dir() || is_skip_listed_dir(&entry.file_name().to_string_lossy()) {
            continue;
        }

        collect_shallow_workspace_candidate(&path, canonical_root, &mut workspaces, manifest_cache);
        collect_shallow_child_workspaces(&path, canonical_root, &mut workspaces, manifest_cache);
    }

    workspaces
}

fn collect_shallow_child_workspaces(
    parent: &Path,
    canonical_root: &Path,
    workspaces: &mut Vec<(WorkspaceInfo, Vec<String>)>,
    manifest_cache: &mut ManifestCache,
) {
    let Ok(child_entries) = std::fs::read_dir(parent) else {
        return;
    };
    for child in child_entries.filter_map(Result::ok) {
        let child_path = child.path();
        if !child_path.is_dir() || is_skip_listed_dir(&child.file_name().to_string_lossy()) {
            continue;
        }

        collect_shallow_workspace_candidate(
            &child_path,
            canonical_root,
            workspaces,
            manifest_cache,
        );
    }
}

fn collect_shallow_workspace_candidate(
    dir: &Path,
    canonical_root: &Path,
    workspaces: &mut Vec<(WorkspaceInfo, Vec<String>)>,
    manifest_cache: &mut ManifestCache,
) {
    let canonical_dir = dunce::canonicalize(dir).unwrap_or_else(|_| dir.to_path_buf());
    if canonical_dir == *canonical_root || !canonical_dir.starts_with(canonical_root) {
        return;
    }

    let Ok(Some((name, _pkg, dep_names))) =
        load_member_package_manifest_cached(dir, manifest_cache)
    else {
        return;
    };

    workspaces.push((
        WorkspaceInfo {
            root: dir.to_path_buf(),
            name,
            is_internal_dependency: false,
        },
        dep_names,
    ));
}

/// Deduplicate workspaces by canonical path and mark internal dependencies.
///
/// Overlapping sources (npm workspaces + tsconfig references pointing to the same
/// directory) are collapsed. npm-discovered entries take precedence (they appear first).
/// Workspaces depended on by other workspaces are marked as `is_internal_dependency`.
fn mark_internal_dependencies(workspaces: &mut Vec<(WorkspaceInfo, Vec<String>)>) {
    {
        let mut seen = rustc_hash::FxHashSet::default();
        workspaces.retain(|(ws, _)| {
            let canonical = dunce::canonicalize(&ws.root).unwrap_or_else(|_| ws.root.clone());
            seen.insert(canonical)
        });
    }

    let all_dep_names: rustc_hash::FxHashSet<String> = workspaces
        .iter()
        .flat_map(|(_, deps)| deps.iter().cloned())
        .collect();

    for (ws, _) in &mut *workspaces {
        ws.is_internal_dependency = all_dep_names.contains(&ws.name);
    }
}

/// Extract the directory name as a string, for workspace name fallback.
fn dir_name(dir: &Path) -> String {
    dir.file_name()
        .map(|n| n.to_string_lossy().to_string())
        .unwrap_or_default()
}

/// Build the issue-2366 repository shape: one glob declared in both
/// `package.json` (spelled `./pkgs/*`) and `pnpm-workspace.yaml` (spelled
/// `pkgs/*`), over two directories that carry no `package.json`.
///
/// Shared with the emission tests in `diagnostics.rs`, which assert that the
/// aggregated stderr warning counts each directory once.
///
/// # Panics
///
/// Panics if a fixture directory or file cannot be written.
#[cfg(test)]
pub fn write_two_manifest_glob_project(root: &Path) {
    std::fs::create_dir_all(root.join("pkgs/aaa")).unwrap();
    std::fs::create_dir_all(root.join("pkgs/bbb")).unwrap();
    std::fs::write(
        root.join("package.json"),
        r#"{"name":"two-manifest-root","private":true,"workspaces":["./pkgs/*"]}"#,
    )
    .unwrap();
    std::fs::write(
        root.join("pnpm-workspace.yaml"),
        "packages:\n  - \"pkgs/*\"\n",
    )
    .unwrap();
    std::fs::write(root.join("pkgs/aaa/readme.txt"), "no package.json here\n").unwrap();
    std::fs::write(root.join("pkgs/bbb/readme.txt"), "no package.json here\n").unwrap();
}

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

    #[test]
    fn public_workspace_selection_supports_exact_names_and_globs() {
        assert!(!workspace_is_public("@scope/core", &[]));
        assert!(workspace_is_public(
            "@scope/core",
            &["@scope/core".to_string()]
        ));
        assert!(workspace_is_public(
            "@scope/core",
            &["@scope/*".to_string()]
        ));
        assert!(!workspace_is_public(
            "@scope/core",
            &["[invalid".to_string(), "@other/*".to_string()]
        ));
    }

    #[test]
    fn discover_workspaces_from_tsconfig_references() {
        let temp_dir = std::env::temp_dir().join("fallow-test-ws-tsconfig-refs");
        let _ = std::fs::remove_dir_all(&temp_dir);
        std::fs::create_dir_all(temp_dir.join("packages/core")).unwrap();
        std::fs::create_dir_all(temp_dir.join("packages/ui")).unwrap();

        std::fs::write(
            temp_dir.join("tsconfig.json"),
            r#"{"references": [{"path": "./packages/core"}, {"path": "./packages/ui"}]}"#,
        )
        .unwrap();

        std::fs::write(
            temp_dir.join("packages/core/package.json"),
            r#"{"name": "@project/core"}"#,
        )
        .unwrap();

        let workspaces = discover_workspaces(&temp_dir);
        assert_eq!(workspaces.len(), 2);
        assert!(workspaces.iter().any(|ws| ws.name == "@project/core"));
        assert!(workspaces.iter().any(|ws| ws.name == "ui"));

        let _ = std::fs::remove_dir_all(&temp_dir);
    }

    #[test]
    fn tsconfig_references_outside_root_rejected() {
        let temp_dir = std::env::temp_dir().join("fallow-test-tsconfig-outside");
        let _ = std::fs::remove_dir_all(&temp_dir);
        std::fs::create_dir_all(temp_dir.join("project/packages/core")).unwrap();
        std::fs::create_dir_all(temp_dir.join("outside")).unwrap();

        std::fs::write(
            temp_dir.join("project/tsconfig.json"),
            r#"{"references": [{"path": "./packages/core"}, {"path": "../outside"}]}"#,
        )
        .unwrap();

        let workspaces = discover_workspaces(&temp_dir.join("project"));
        assert_eq!(
            workspaces.len(),
            1,
            "reference outside project root should be rejected: {workspaces:?}"
        );
        assert!(
            workspaces[0]
                .root
                .to_string_lossy()
                .contains("packages/core")
        );

        let _ = std::fs::remove_dir_all(&temp_dir);
    }

    #[test]
    fn dir_name_extracts_last_component() {
        assert_eq!(dir_name(Path::new("/project/packages/core")), "core");
        assert_eq!(dir_name(Path::new("/my-app")), "my-app");
    }

    #[test]
    fn dir_name_empty_for_root_path() {
        assert_eq!(dir_name(Path::new("/")), "");
    }

    #[test]
    fn workspace_config_deserialize_json() {
        let json = r#"{"patterns": ["packages/*", "apps/*"]}"#;
        let config: WorkspaceConfig = serde_json::from_str(json).unwrap();
        assert_eq!(config.patterns, vec!["packages/*", "apps/*"]);
    }

    #[test]
    fn workspace_config_deserialize_empty_patterns() {
        let json = r#"{"patterns": []}"#;
        let config: WorkspaceConfig = serde_json::from_str(json).unwrap();
        assert!(config.patterns.is_empty());
    }

    #[test]
    fn workspace_config_default_patterns() {
        let json = "{}";
        let config: WorkspaceConfig = serde_json::from_str(json).unwrap();
        assert!(config.patterns.is_empty());
    }

    #[test]
    fn workspace_info_default_not_internal() {
        let ws = WorkspaceInfo {
            root: PathBuf::from("/project/packages/a"),
            name: "a".to_string(),
            is_internal_dependency: false,
        };
        assert!(!ws.is_internal_dependency);
    }

    #[test]
    fn mark_internal_deps_detects_cross_references() {
        let temp_dir = tempfile::tempdir().expect("create temp dir");
        let pkg_a = temp_dir.path().join("a");
        let pkg_b = temp_dir.path().join("b");
        std::fs::create_dir_all(&pkg_a).unwrap();
        std::fs::create_dir_all(&pkg_b).unwrap();

        let mut workspaces = vec![
            (
                WorkspaceInfo {
                    root: pkg_a,
                    name: "@scope/a".to_string(),
                    is_internal_dependency: false,
                },
                vec!["@scope/b".to_string()], // "a" depends on "b"
            ),
            (
                WorkspaceInfo {
                    root: pkg_b,
                    name: "@scope/b".to_string(),
                    is_internal_dependency: false,
                },
                vec!["lodash".to_string()], // "b" depends on external only
            ),
        ];

        mark_internal_dependencies(&mut workspaces);

        let ws_a = workspaces
            .iter()
            .find(|(ws, _)| ws.name == "@scope/a")
            .unwrap();
        assert!(
            !ws_a.0.is_internal_dependency,
            "a is not depended on by others"
        );

        let ws_b = workspaces
            .iter()
            .find(|(ws, _)| ws.name == "@scope/b")
            .unwrap();
        assert!(ws_b.0.is_internal_dependency, "b is depended on by a");
    }

    #[test]
    fn mark_internal_deps_no_cross_references() {
        let temp_dir = tempfile::tempdir().expect("create temp dir");
        let pkg_a = temp_dir.path().join("a");
        let pkg_b = temp_dir.path().join("b");
        std::fs::create_dir_all(&pkg_a).unwrap();
        std::fs::create_dir_all(&pkg_b).unwrap();

        let mut workspaces = vec![
            (
                WorkspaceInfo {
                    root: pkg_a,
                    name: "a".to_string(),
                    is_internal_dependency: false,
                },
                vec!["react".to_string()],
            ),
            (
                WorkspaceInfo {
                    root: pkg_b,
                    name: "b".to_string(),
                    is_internal_dependency: false,
                },
                vec!["lodash".to_string()],
            ),
        ];

        mark_internal_dependencies(&mut workspaces);

        assert!(!workspaces[0].0.is_internal_dependency);
        assert!(!workspaces[1].0.is_internal_dependency);
    }

    #[test]
    fn mark_internal_deps_deduplicates_by_path() {
        let temp_dir = tempfile::tempdir().expect("create temp dir");
        let pkg_a = temp_dir.path().join("a");
        std::fs::create_dir_all(&pkg_a).unwrap();

        let mut workspaces = vec![
            (
                WorkspaceInfo {
                    root: pkg_a.clone(),
                    name: "a".to_string(),
                    is_internal_dependency: false,
                },
                vec![],
            ),
            (
                WorkspaceInfo {
                    root: pkg_a,
                    name: "a".to_string(),
                    is_internal_dependency: false,
                },
                vec![],
            ),
        ];

        mark_internal_dependencies(&mut workspaces);
        assert_eq!(
            workspaces.len(),
            1,
            "duplicate paths should be deduplicated"
        );
    }

    #[test]
    fn collect_patterns_from_package_json() {
        let dir = tempfile::tempdir().expect("create temp dir");
        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/*", "apps/*"]}"#,
        )
        .unwrap();

        let patterns = collect_workspace_patterns(dir.path()).expect("valid root package.json");
        assert_eq!(patterns, vec!["packages/*", "apps/*"]);
    }

    #[test]
    fn collect_patterns_from_pnpm_workspace() {
        let dir = tempfile::tempdir().expect("create temp dir");
        std::fs::write(
            dir.path().join("pnpm-workspace.yaml"),
            "packages:\n  - 'packages/*'\n  - 'libs/*'\n",
        )
        .unwrap();

        let patterns = collect_workspace_patterns(dir.path()).expect("no root package.json");
        assert_eq!(patterns, vec!["packages/*", "libs/*"]);
    }

    #[test]
    fn collect_patterns_combines_sources() {
        let dir = tempfile::tempdir().expect("create temp dir");
        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/*"]}"#,
        )
        .unwrap();
        std::fs::write(
            dir.path().join("pnpm-workspace.yaml"),
            "packages:\n  - 'apps/*'\n",
        )
        .unwrap();

        let patterns = collect_workspace_patterns(dir.path()).expect("valid root package.json");
        assert!(patterns.contains(&"packages/*".to_string()));
        assert!(patterns.contains(&"apps/*".to_string()));
    }

    #[test]
    fn collect_patterns_empty_when_no_configs() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let patterns = collect_workspace_patterns(dir.path()).expect("no root package.json");
        assert!(patterns.is_empty());
    }

    #[test]
    fn discover_workspaces_from_package_json() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let pkg_a = dir.path().join("packages").join("a");
        let pkg_b = dir.path().join("packages").join("b");
        std::fs::create_dir_all(&pkg_a).unwrap();
        std::fs::create_dir_all(&pkg_b).unwrap();

        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/*"]}"#,
        )
        .unwrap();
        std::fs::write(
            pkg_a.join("package.json"),
            r#"{"name": "@test/a", "dependencies": {"@test/b": "workspace:*"}}"#,
        )
        .unwrap();
        std::fs::write(pkg_b.join("package.json"), r#"{"name": "@test/b"}"#).unwrap();

        let workspaces = discover_workspaces(dir.path());
        assert_eq!(workspaces.len(), 2);

        let ws_a = workspaces.iter().find(|ws| ws.name == "@test/a").unwrap();
        assert!(!ws_a.is_internal_dependency);

        let ws_b = workspaces.iter().find(|ws| ws.name == "@test/b").unwrap();
        assert!(ws_b.is_internal_dependency, "b is depended on by a");
    }

    #[test]
    fn discover_workspaces_empty_project() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let workspaces = discover_workspaces(dir.path());
        assert!(workspaces.is_empty());
    }

    #[test]
    fn discover_workspaces_falls_back_to_shallow_packages_without_workspace_config() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let benchmarks = dir.path().join("benchmarks");
        let vscode = dir.path().join("editors").join("vscode");
        let deep = dir.path().join("tests").join("fixtures").join("demo");
        std::fs::create_dir_all(&benchmarks).unwrap();
        std::fs::create_dir_all(&vscode).unwrap();
        std::fs::create_dir_all(&deep).unwrap();

        std::fs::write(benchmarks.join("package.json"), r#"{"name": "benchmarks"}"#).unwrap();
        std::fs::write(vscode.join("package.json"), r#"{"name": "fallow-vscode"}"#).unwrap();
        std::fs::write(deep.join("package.json"), r#"{"name": "deep-fixture"}"#).unwrap();

        let workspaces = discover_workspaces(dir.path());
        let names: Vec<&str> = workspaces.iter().map(|ws| ws.name.as_str()).collect();

        assert!(
            names.contains(&"benchmarks"),
            "top-level nested package should be discovered: {workspaces:?}"
        );
        assert!(
            names.contains(&"fallow-vscode"),
            "second-level nested package should be discovered: {workspaces:?}"
        );
        assert!(
            !names.contains(&"deep-fixture"),
            "fallback should stay shallow and skip deep fixtures: {workspaces:?}"
        );
    }

    #[test]
    fn discover_workspaces_with_negated_patterns() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let pkg_a = dir.path().join("packages").join("a");
        let pkg_test = dir.path().join("packages").join("test-utils");
        std::fs::create_dir_all(&pkg_a).unwrap();
        std::fs::create_dir_all(&pkg_test).unwrap();

        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/*", "!packages/test-*"]}"#,
        )
        .unwrap();
        std::fs::write(pkg_a.join("package.json"), r#"{"name": "a"}"#).unwrap();
        std::fs::write(pkg_test.join("package.json"), r#"{"name": "test-utils"}"#).unwrap();

        let workspaces = discover_workspaces(dir.path());
        assert_eq!(workspaces.len(), 1);
        assert_eq!(workspaces[0].name, "a");
    }

    #[test]
    fn discover_workspaces_skips_root_as_workspace() {
        let dir = tempfile::tempdir().expect("create temp dir");
        std::fs::write(
            dir.path().join("pnpm-workspace.yaml"),
            "packages:\n  - '.'\n",
        )
        .unwrap();
        std::fs::write(dir.path().join("package.json"), r#"{"name": "root"}"#).unwrap();

        let workspaces = discover_workspaces(dir.path());
        assert!(
            workspaces.is_empty(),
            "root directory should not be added as workspace"
        );
    }

    #[test]
    fn discover_workspaces_name_fallback_to_dir_name() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let pkg_a = dir.path().join("packages").join("my-app");
        std::fs::create_dir_all(&pkg_a).unwrap();

        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/*"]}"#,
        )
        .unwrap();
        std::fs::write(pkg_a.join("package.json"), "{}").unwrap();

        let workspaces = discover_workspaces(dir.path());
        assert_eq!(workspaces.len(), 1);
        assert_eq!(workspaces[0].name, "my-app", "should fall back to dir name");
    }

    #[test]
    fn discover_workspaces_explicit_patterns_disable_shallow_fallback() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let pkg_a = dir.path().join("packages").join("a");
        let benchmarks = dir.path().join("benchmarks");
        std::fs::create_dir_all(&pkg_a).unwrap();
        std::fs::create_dir_all(&benchmarks).unwrap();

        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/*"]}"#,
        )
        .unwrap();
        std::fs::write(pkg_a.join("package.json"), r#"{"name": "a"}"#).unwrap();
        std::fs::write(benchmarks.join("package.json"), r#"{"name": "benchmarks"}"#).unwrap();

        let workspaces = discover_workspaces(dir.path());
        let names: Vec<&str> = workspaces.iter().map(|ws| ws.name.as_str()).collect();

        assert_eq!(workspaces.len(), 1);
        assert!(names.contains(&"a"));
        assert!(
            !names.contains(&"benchmarks"),
            "explicit workspace config should keep undeclared packages out: {workspaces:?}"
        );
    }

    #[test]
    fn discover_workspaces_recovers_package_under_bare_glob_intermediate() {
        // Issue #842 (reporter metrists/metrists): root declares
        // `["./packages/*", "./themes/*"]`, but the real package lives two levels
        // deep at `packages/themes/metrists-theme-next` while `packages/themes`
        // itself has no package.json. The single-level glob only matches the bare
        // `packages/themes`; without recovery the deep package is never discovered,
        // its files fall back to the root manifest, and its declared deps (react)
        // are reported as unlisted. Discovery must recover the named deep package.
        let dir = tempfile::tempdir().expect("create temp dir");
        let theme = dir
            .path()
            .join("packages")
            .join("themes")
            .join("metrists-theme-next");
        std::fs::create_dir_all(&theme).unwrap();
        std::fs::write(
            dir.path().join("package.json"),
            r#"{"name": "metrists-monorepo", "workspaces": ["./packages/*", "./themes/*"]}"#,
        )
        .unwrap();
        // packages/themes intentionally has NO package.json (bare grouping dir).
        std::fs::write(
            theme.join("package.json"),
            r#"{"name": "metrists-theme-next", "dependencies": {"react": "^18"}}"#,
        )
        .unwrap();

        let workspaces = discover_workspaces(dir.path());
        assert!(
            workspaces.iter().any(|ws| ws.name == "metrists-theme-next"),
            "deep package under a bare glob-matched intermediate must be discovered: {workspaces:?}"
        );
    }

    fn write_package(dir: &Path, manifest: &str) {
        std::fs::create_dir_all(dir).unwrap();
        std::fs::write(dir.join("package.json"), manifest).unwrap();
    }

    #[test]
    fn discover_workspaces_follows_deep_link_dependency_targets() {
        // A yarn-era monorepo (for example older Kibana) has no `workspaces`
        // field. The root lists each package as a `link:` or `file:`
        // dependency, three to five directories deep.
        let dir = tempfile::tempdir().expect("create temp dir");
        let root = dir.path();
        std::fs::write(
            root.join("package.json"),
            r#"{
              "name": "kibana",
              "dependencies": {
                "@kbn/foo": "link:src/platform/packages/shared/kbn-foo",
                "@kbn/app": "link:./x-pack/solutions/search/plugins/app"
              },
              "devDependencies": {
                "@kbn/test": "file:src/platform/packages/private/kbn-test"
              }
            }"#,
        )
        .unwrap();
        write_package(
            &root.join("src/platform/packages/shared/kbn-foo"),
            r#"{"name": "@kbn/foo"}"#,
        );
        write_package(
            &root.join("x-pack/solutions/search/plugins/app"),
            r#"{"name": "@kbn/app", "dependencies": {"@kbn/foo": "link:../../../../../src/platform/packages/shared/kbn-foo"}}"#,
        );
        write_package(
            &root.join("src/platform/packages/private/kbn-test"),
            r#"{"name": "@kbn/test"}"#,
        );

        let workspaces = discover_workspaces(root);
        let mut names: Vec<&str> = workspaces.iter().map(|ws| ws.name.as_str()).collect();
        names.sort_unstable();
        assert_eq!(names, ["@kbn/app", "@kbn/foo", "@kbn/test"]);
        let foo = workspaces.iter().find(|ws| ws.name == "@kbn/foo").unwrap();
        assert!(foo.is_internal_dependency, "{workspaces:?}");
    }

    #[test]
    fn discover_workspaces_skips_link_targets_outside_root_or_without_manifest() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let root = dir.path().join("repo");
        std::fs::create_dir_all(&root).unwrap();
        write_package(&dir.path().join("sibling"), r#"{"name": "sibling"}"#);
        std::fs::create_dir_all(root.join("packages/no-manifest")).unwrap();
        std::fs::write(root.join("vendor.tgz"), "").unwrap();
        std::fs::write(
            root.join("package.json"),
            r#"{
              "dependencies": {
                "sibling": "link:../sibling",
                "absolute": "file:/definitely/not/here",
                "missing": "link:packages/missing",
                "no-manifest": "link:packages/no-manifest",
                "tarball": "file:./vendor.tgz",
                "self": "link:.",
                "react": "^18.0.0"
              }
            }"#,
        )
        .unwrap();

        let workspaces = discover_workspaces(&root);
        assert!(workspaces.is_empty(), "{workspaces:?}");
    }

    #[test]
    fn discover_workspaces_skips_link_targets_that_source_discovery_skips() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let root = dir.path();
        std::fs::write(
            root.join("package.json"),
            r#"{
              "dependencies": {
                "ylib": "file:.yalc/ylib",
                "nm": "link:node_modules/nm",
                "blib": "link:build/blib",
                "dlib": "link:dist/dlib",
                "clib": "link:packages/coverage/clib",
                "vlib": "link:vendor/deep/vlib",
                "kept": "link:libs/deep/kept"
              }
            }"#,
        )
        .unwrap();
        for (path, name) in [
            (".yalc/ylib", "ylib"),
            ("node_modules/nm", "nm"),
            ("build/blib", "blib"),
            ("dist/dlib", "dlib"),
            ("packages/coverage/clib", "clib"),
            ("vendor/deep/vlib", "vlib"),
            ("libs/deep/kept", "kept"),
        ] {
            write_package(&root.join(path), &format!(r#"{{"name": "{name}"}}"#));
        }
        let mut builder = globset::GlobSetBuilder::new();
        builder.add(globset::Glob::new("vendor/**").unwrap());
        let ignore = crate::IgnorePatternSet::from(builder.build().unwrap());

        let (workspaces, _) = discover_workspaces_with_diagnostics(root, &ignore).unwrap();
        let names: Vec<&str> = workspaces.iter().map(|ws| ws.name.as_str()).collect();
        assert_eq!(names, ["kept"]);
    }

    fn write_gitignored_link_fixture(root: &Path) {
        std::fs::write(
            root.join("package.json"),
            r#"{
              "dependencies": {
                "vlib": "file:vendor/deep/vlib",
                "plib": "link:libs/private/plib",
                "kept": "link:libs/deep/kept"
              }
            }"#,
        )
        .unwrap();
        for (path, name) in [
            ("vendor/deep/vlib", "vlib"),
            ("libs/private/plib", "plib"),
            ("libs/deep/kept", "kept"),
        ] {
            write_package(&root.join(path), &format!(r#"{{"name": "{name}"}}"#));
        }
        std::fs::write(root.join(".gitignore"), "vendor/\n").unwrap();
        std::fs::write(root.join("libs/.gitignore"), "private/\n").unwrap();
    }

    fn link_workspace_names(root: &Path) -> Vec<String> {
        let ignore = crate::IgnorePatternSet::default();
        let (workspaces, _) = discover_workspaces_with_diagnostics(root, &ignore).unwrap();
        let mut names: Vec<String> = workspaces.into_iter().map(|ws| ws.name).collect();
        names.sort_unstable();
        names
    }

    #[test]
    fn discover_workspaces_skips_gitignored_link_targets() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let root = dir.path();
        std::fs::create_dir_all(root.join(".git")).unwrap();
        write_gitignored_link_fixture(root);

        assert_eq!(link_workspace_names(root), ["kept"]);
    }

    #[test]
    fn discover_workspaces_keeps_link_targets_when_gitignore_does_not_apply() {
        // Source discovery honors `.gitignore` only inside a git repository.
        let dir = tempfile::tempdir().expect("create temp dir");
        let root = dir.path();
        write_gitignored_link_fixture(root);

        assert_eq!(link_workspace_names(root), ["kept", "plib", "vlib"]);
    }

    #[test]
    fn discover_workspaces_dedupes_link_targets_with_declared_workspaces() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let root = dir.path();
        std::fs::write(
            root.join("package.json"),
            r#"{
              "workspaces": ["packages/*"],
              "dependencies": {
                "a": "link:packages/a",
                "tool": "link:tools/deep/tool"
              }
            }"#,
        )
        .unwrap();
        write_package(&root.join("packages/a"), r#"{"name": "a"}"#);
        write_package(&root.join("tools/deep/tool"), r#"{"name": "tool"}"#);

        let workspaces = discover_workspaces(root);
        let mut names: Vec<&str> = workspaces.iter().map(|ws| ws.name.as_str()).collect();
        names.sort_unstable();
        assert_eq!(names, ["a", "tool"]);
        assert!(
            find_undeclared_workspaces(root, &workspaces).is_empty(),
            "a linked package is declared, not undeclared"
        );
    }

    #[test]
    fn undeclared_workspace_detected() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let pkg_a = dir.path().join("packages").join("a");
        let pkg_b = dir.path().join("packages").join("b");
        std::fs::create_dir_all(&pkg_a).unwrap();
        std::fs::create_dir_all(&pkg_b).unwrap();

        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/a"]}"#,
        )
        .unwrap();
        std::fs::write(pkg_a.join("package.json"), r#"{"name": "a"}"#).unwrap();
        std::fs::write(pkg_b.join("package.json"), r#"{"name": "b"}"#).unwrap();

        let declared = discover_workspaces(dir.path());
        assert_eq!(declared.len(), 1);

        let undeclared = find_undeclared_workspaces(dir.path(), &declared);
        assert_eq!(undeclared.len(), 1);
        assert!(
            undeclared[0]
                .path
                .to_string_lossy()
                .replace('\\', "/")
                .contains("packages/b"),
            "should detect packages/b as undeclared: {:?}",
            undeclared[0].path
        );
    }

    #[test]
    fn no_undeclared_when_all_covered() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let pkg_a = dir.path().join("packages").join("a");
        std::fs::create_dir_all(&pkg_a).unwrap();

        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/*"]}"#,
        )
        .unwrap();
        std::fs::write(pkg_a.join("package.json"), r#"{"name": "a"}"#).unwrap();

        let declared = discover_workspaces(dir.path());
        let undeclared = find_undeclared_workspaces(dir.path(), &declared);
        assert!(undeclared.is_empty());
    }

    #[test]
    fn no_undeclared_when_no_workspace_patterns() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let sub = dir.path().join("lib");
        std::fs::create_dir_all(&sub).unwrap();

        std::fs::write(dir.path().join("package.json"), r#"{"name": "app"}"#).unwrap();
        std::fs::write(sub.join("package.json"), r#"{"name": "lib"}"#).unwrap();

        let undeclared = find_undeclared_workspaces(dir.path(), &[]);
        assert!(
            undeclared.is_empty(),
            "should skip check when no workspace patterns exist"
        );
    }

    #[test]
    fn undeclared_skips_node_modules_and_hidden_dirs() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let nm = dir.path().join("node_modules").join("some-pkg");
        let hidden = dir.path().join(".hidden");
        std::fs::create_dir_all(&nm).unwrap();
        std::fs::create_dir_all(&hidden).unwrap();

        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/*"]}"#,
        )
        .unwrap();
        std::fs::write(nm.join("package.json"), r#"{"name": "nm-pkg"}"#).unwrap();
        std::fs::write(hidden.join("package.json"), r#"{"name": "hidden"}"#).unwrap();

        let undeclared = find_undeclared_workspaces(dir.path(), &[]);
        assert!(
            undeclared.is_empty(),
            "should not flag node_modules or hidden directories"
        );
    }

    fn build_globset(patterns: &[&str]) -> crate::IgnorePatternSet {
        let mut builder = globset::GlobSetBuilder::new();
        for pattern in patterns {
            builder.add(globset::Glob::new(pattern).expect("valid glob"));
        }
        crate::IgnorePatternSet::from(builder.build().expect("build globset"))
    }

    #[test]
    fn undeclared_skips_dirs_matching_ignore_patterns() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let pkg_a = dir.path().join("packages").join("a");
        let vitest_ref = dir.path().join("references").join("vitest");
        let tanstack_ref = dir.path().join("references").join("tanstack-router");
        std::fs::create_dir_all(&pkg_a).unwrap();
        std::fs::create_dir_all(&vitest_ref).unwrap();
        std::fs::create_dir_all(&tanstack_ref).unwrap();

        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/*"]}"#,
        )
        .unwrap();
        std::fs::write(pkg_a.join("package.json"), r#"{"name": "a"}"#).unwrap();
        std::fs::write(
            vitest_ref.join("package.json"),
            r#"{"name": "vitest-reference"}"#,
        )
        .unwrap();
        std::fs::write(
            tanstack_ref.join("package.json"),
            r#"{"name": "tanstack-reference"}"#,
        )
        .unwrap();

        let declared = discover_workspaces(dir.path());
        let ignore = build_globset(&["references/*"]);
        let undeclared = find_undeclared_workspaces_with_ignores(dir.path(), &declared, &ignore);
        assert!(
            undeclared.is_empty(),
            "references/* should be ignored: {undeclared:?}"
        );
    }

    #[test]
    fn undeclared_still_reported_when_ignore_does_not_match() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let pkg_b = dir.path().join("packages").join("b");
        std::fs::create_dir_all(&pkg_b).unwrap();

        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/a"]}"#,
        )
        .unwrap();
        std::fs::write(pkg_b.join("package.json"), r#"{"name": "b"}"#).unwrap();

        let declared = discover_workspaces(dir.path());
        let ignore = build_globset(&["references/*"]);
        let undeclared = find_undeclared_workspaces_with_ignores(dir.path(), &declared, &ignore);
        assert_eq!(
            undeclared.len(),
            1,
            "non-matching ignore patterns should not silence other undeclared dirs"
        );
    }

    #[test]
    fn undeclared_skips_dirs_matching_package_json_glob() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let pkg_a = dir.path().join("packages").join("a");
        let vitest_ref = dir.path().join("references").join("vitest");
        std::fs::create_dir_all(&pkg_a).unwrap();
        std::fs::create_dir_all(&vitest_ref).unwrap();

        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/*"]}"#,
        )
        .unwrap();
        std::fs::write(pkg_a.join("package.json"), r#"{"name": "a"}"#).unwrap();
        std::fs::write(
            vitest_ref.join("package.json"),
            r#"{"name": "vitest-reference"}"#,
        )
        .unwrap();

        let declared = discover_workspaces(dir.path());
        let ignore = build_globset(&["references/*/package.json"]);
        let undeclared = find_undeclared_workspaces_with_ignores(dir.path(), &declared, &ignore);
        assert!(
            undeclared.is_empty(),
            "package.json-suffixed glob should silence the warning: {undeclared:?}"
        );
    }

    #[test]
    fn undeclared_skips_dirs_matching_doublestar_ignore() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let pkg_a = dir.path().join("packages").join("a");
        let nested_ref = dir.path().join("references").join("vitest");
        std::fs::create_dir_all(&pkg_a).unwrap();
        std::fs::create_dir_all(&nested_ref).unwrap();

        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/*"]}"#,
        )
        .unwrap();
        std::fs::write(pkg_a.join("package.json"), r#"{"name": "a"}"#).unwrap();
        std::fs::write(
            nested_ref.join("package.json"),
            r#"{"name": "vitest-reference"}"#,
        )
        .unwrap();

        let declared = discover_workspaces(dir.path());
        let ignore = build_globset(&["**/references/**"]);
        let undeclared = find_undeclared_workspaces_with_ignores(dir.path(), &declared, &ignore);
        assert!(
            undeclared.is_empty(),
            "**/references/** should ignore nested package.json dirs: {undeclared:?}"
        );
    }

    #[test]
    fn malformed_workspace_package_json_emits_diagnostic() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let pkg_a = dir.path().join("packages").join("a");
        let pkg_bad = dir.path().join("packages").join("bad");
        std::fs::create_dir_all(&pkg_a).unwrap();
        std::fs::create_dir_all(&pkg_bad).unwrap();
        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/*"]}"#,
        )
        .unwrap();
        std::fs::write(pkg_a.join("package.json"), r#"{"name": "a"}"#).unwrap();
        std::fs::write(pkg_bad.join("package.json"), r#"{"name": "bad",}"#).unwrap();

        let (result, captured) = capture_workspace_warnings(|| {
            discover_workspaces_with_diagnostics(dir.path(), &crate::IgnorePatternSet::empty())
        });
        let (workspaces, diagnostics) = result.expect("root package.json is valid");

        assert_eq!(workspaces.len(), 1, "the valid workspace still discovers");
        assert_eq!(workspaces[0].name, "a");
        assert_eq!(diagnostics.len(), 1);
        assert!(matches!(
            diagnostics[0].kind,
            WorkspaceDiagnosticKind::MalformedPackageJson { .. }
        ));
        assert!(
            captured
                .iter()
                .any(|d| matches!(d.kind, WorkspaceDiagnosticKind::MalformedPackageJson { .. }))
        );
    }

    #[test]
    fn malformed_bridge_deno_config_emits_recoverable_member_diagnostic() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let pkg_good = dir.path().join("packages/good");
        let pkg_bad = dir.path().join("packages/bad");
        std::fs::create_dir_all(&pkg_good).unwrap();
        std::fs::create_dir_all(&pkg_bad).unwrap();
        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/*"]}"#,
        )
        .unwrap();
        std::fs::write(pkg_good.join("package.json"), r#"{"name": "good"}"#).unwrap();
        std::fs::write(pkg_bad.join("package.json"), r#"{"name": "bad"}"#).unwrap();
        std::fs::write(pkg_bad.join("deno.jsonc"), "{ imports: [ }").unwrap();

        let result =
            discover_workspaces_with_diagnostics(dir.path(), &crate::IgnorePatternSet::empty());
        let (workspaces, diagnostics) = result.expect("root package.json is valid");

        assert_eq!(workspaces.len(), 1, "valid sibling should still discover");
        assert_eq!(workspaces[0].name, "good");
        assert_eq!(diagnostics.len(), 1);
        assert!(matches!(
            diagnostics[0].kind,
            WorkspaceDiagnosticKind::MalformedPackageJson { .. }
        ));
        assert!(diagnostics[0].message.contains("deno.jsonc"));
    }

    /// A member reached through two workspace sources (npm glob + tsconfig
    /// reference) is diagnosed, and diagnosed ONCE: the two sources produce
    /// byte-identical diagnostics, so the discovery fold reports the member
    /// once where every envelope used to report it twice (issue #2366).
    ///
    /// That the tsconfig leg replayed the memoized `Err` rather than silently
    /// resolving is NOT observable here (both `Ok(None)` and `Err` fall back
    /// to the directory name, and the fold hides a second diagnostic);
    /// [`manifest_cache_replays_the_same_err_on_every_visit`] pins it.
    #[test]
    fn malformed_member_reached_via_two_sources_is_diagnosed_once() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let pkg_good = dir.path().join("packages").join("good");
        let pkg_bad = dir.path().join("packages").join("bad");
        std::fs::create_dir_all(&pkg_good).unwrap();
        std::fs::create_dir_all(&pkg_bad).unwrap();
        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/*"]}"#,
        )
        .unwrap();
        std::fs::write(
            dir.path().join("tsconfig.json"),
            r#"{"references": [{"path": "./packages/bad"}]}"#,
        )
        .unwrap();
        std::fs::write(pkg_good.join("package.json"), r#"{"name": "good"}"#).unwrap();
        std::fs::write(pkg_bad.join("package.json"), r"{,}").unwrap();

        let (result, _) = capture_workspace_warnings(|| {
            discover_workspaces_with_diagnostics(dir.path(), &crate::IgnorePatternSet::empty())
        });
        let (workspaces, diagnostics) = result.expect("root package.json is valid");

        assert!(
            workspaces.iter().any(|w| w.name == "good"),
            "valid sibling should still discover: {workspaces:?}"
        );
        assert!(
            workspaces.iter().any(|w| w.name == "bad"),
            "tsconfig reference falls back to the directory name on manifest error"
        );
        let malformed: Vec<_> = diagnostics
            .iter()
            .filter(|d| matches!(d.kind, WorkspaceDiagnosticKind::MalformedPackageJson { .. }))
            .collect();
        assert_eq!(
            malformed.len(),
            1,
            "the replayed Err is one diagnostic, not one per source: {diagnostics:?}"
        );
        assert!(
            malformed[0].path.ends_with("bad"),
            "the diagnostic points at the malformed member"
        );
    }

    /// The per-discovery [`ManifestCache`] memoizes `Err` outcomes and every
    /// repeat visit must get the same `Err` back. Both callers that can reach
    /// one member ([`register_matched_workspace`] for an npm glob and
    /// [`load_tsconfig_workspace_package`] for a tsconfig reference) diagnose
    /// from the outcome the cache hands them, so a replay that degraded to
    /// `Ok(None)` would silently drop the diagnostic on whichever source is
    /// second. No end-to-end assertion can witness that: `Ok(None)` and `Err`
    /// both fall back to the directory name for the workspace, and the
    /// discovery fold collapses the two byte-identical diagnostics into one
    /// either way.
    #[test]
    fn manifest_cache_replays_the_same_err_on_every_visit() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let member = dir.path().join("packages").join("bad");
        std::fs::create_dir_all(&member).unwrap();
        std::fs::write(member.join("package.json"), r"{,}").unwrap();

        let mut cache = ManifestCache::default();
        let first = load_member_package_manifest_cached(&member, &mut cache)
            .expect_err("a malformed member manifest is an Err");
        assert_eq!(cache.len(), 1, "the outcome is memoized: {cache:?}");

        std::fs::remove_file(member.join("package.json")).unwrap();
        let replayed = load_member_package_manifest_cached(&member, &mut cache)
            .expect_err("the memoized Err replays instead of re-reading the directory");
        assert_eq!(
            replayed, first,
            "the replayed outcome is byte-identical to the first, so both \
             workspace sources diagnose the same member the same way"
        );
    }

    #[test]
    fn multiple_malformed_workspace_package_jsons_all_diagnosed() {
        let dir = tempfile::tempdir().expect("create temp dir");
        for name in ["a", "b", "c"] {
            let pkg = dir.path().join("packages").join(name);
            std::fs::create_dir_all(&pkg).unwrap();
            std::fs::write(pkg.join("package.json"), r"{,}").unwrap();
        }
        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/*"]}"#,
        )
        .unwrap();

        let (result, _) = capture_workspace_warnings(|| {
            discover_workspaces_with_diagnostics(dir.path(), &crate::IgnorePatternSet::empty())
        });
        let (workspaces, diagnostics) = result.expect("root package.json is valid");

        assert!(workspaces.is_empty(), "all three malformed; nothing valid");
        assert_eq!(diagnostics.len(), 3, "each malformed workspace surfaces");
        assert!(
            diagnostics
                .iter()
                .all(|d| matches!(d.kind, WorkspaceDiagnosticKind::MalformedPackageJson { .. })),
            "every diagnostic should be malformed-package-json"
        );
    }

    #[test]
    fn malformed_root_package_json_returns_load_error() {
        let dir = tempfile::tempdir().expect("create temp dir");
        std::fs::write(dir.path().join("package.json"), "this is not json").unwrap();

        let result =
            discover_workspaces_with_diagnostics(dir.path(), &crate::IgnorePatternSet::empty());

        match result {
            Err(WorkspaceLoadError::MalformedRootPackageJson { path, error }) => {
                assert!(path.ends_with("package.json"));
                assert!(!error.is_empty(), "underlying parse error is preserved");
            }
            other => panic!("expected MalformedRootPackageJson, got {other:?}"),
        }
    }

    #[test]
    fn glob_match_without_package_json_emits_diagnostic_unless_skip_listed() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let pkg_a = dir.path().join("packages").join("a");
        let cache_dir = dir.path().join("packages").join(".cache");
        let scratch_dir = dir.path().join("packages").join("scratch");
        std::fs::create_dir_all(&pkg_a).unwrap();
        std::fs::create_dir_all(&cache_dir).unwrap();
        std::fs::create_dir_all(&scratch_dir).unwrap();
        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/*"]}"#,
        )
        .unwrap();
        std::fs::write(pkg_a.join("package.json"), r#"{"name": "a"}"#).unwrap();

        let result =
            discover_workspaces_with_diagnostics(dir.path(), &crate::IgnorePatternSet::empty());
        let (workspaces, diagnostics) = result.expect("root package.json is valid");

        assert_eq!(workspaces.len(), 1);
        let kinds: Vec<&str> = diagnostics.iter().map(|d| d.kind.id()).collect();
        assert!(
            kinds.contains(&"glob-matched-no-package-json"),
            "scratch should diagnose: {kinds:?}"
        );
        assert!(
            !diagnostics.iter().any(|d| d.path.ends_with(".cache")),
            ".cache must be skip-listed: {diagnostics:?}"
        );
    }

    #[test]
    fn glob_match_without_package_json_honors_ignore_patterns() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let pkg_a = dir.path().join("packages").join("a");
        let legacy_dir = dir.path().join("packages").join("legacy");
        std::fs::create_dir_all(&pkg_a).unwrap();
        std::fs::create_dir_all(&legacy_dir).unwrap();
        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/*"]}"#,
        )
        .unwrap();
        std::fs::write(pkg_a.join("package.json"), r#"{"name": "a"}"#).unwrap();

        let mut builder = globset::GlobSetBuilder::new();
        builder.add(globset::Glob::new("packages/legacy").unwrap());
        let ignore = crate::IgnorePatternSet::from(builder.build().unwrap());

        let result = discover_workspaces_with_diagnostics(dir.path(), &ignore);
        let (workspaces, diagnostics) = result.expect("root package.json is valid");

        assert_eq!(workspaces.len(), 1);
        assert!(
            diagnostics.is_empty(),
            "user-excluded path must not produce a diagnostic: {diagnostics:?}"
        );
    }

    #[test]
    fn malformed_tsconfig_emits_diagnostic() {
        let dir = tempfile::tempdir().expect("create temp dir");
        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/*"]}"#,
        )
        .unwrap();
        std::fs::write(dir.path().join("tsconfig.json"), r#"{"references": [,,,]}"#).unwrap();

        let result =
            discover_workspaces_with_diagnostics(dir.path(), &crate::IgnorePatternSet::empty());
        let (_, diagnostics) = result.expect("root package.json is valid");

        assert!(
            diagnostics
                .iter()
                .any(|d| matches!(d.kind, WorkspaceDiagnosticKind::MalformedTsconfig { .. })),
            "expected MalformedTsconfig diagnostic; got: {diagnostics:?}"
        );
    }

    #[test]
    fn tsconfig_missing_reference_dir_emits_diagnostic() {
        let dir = tempfile::tempdir().expect("create temp dir");
        std::fs::write(
            dir.path().join("tsconfig.json"),
            r#"{"references": [{"path": "./packages/missing"}]}"#,
        )
        .unwrap();

        let result =
            discover_workspaces_with_diagnostics(dir.path(), &crate::IgnorePatternSet::empty());
        let (_, diagnostics) = result.expect("no package.json at root is OK");

        assert!(
            diagnostics
                .iter()
                .any(|d| matches!(d.kind, WorkspaceDiagnosticKind::TsconfigReferenceDirMissing)),
            "expected TsconfigReferenceDirMissing; got: {diagnostics:?}"
        );
    }

    #[test]
    fn missing_tsconfig_is_silent() {
        let dir = tempfile::tempdir().expect("create temp dir");

        let result =
            discover_workspaces_with_diagnostics(dir.path(), &crate::IgnorePatternSet::empty());
        let (_, diagnostics) = result.expect("no root package.json is OK");

        assert!(
            !diagnostics
                .iter()
                .any(|d| matches!(d.kind, WorkspaceDiagnosticKind::MalformedTsconfig { .. })),
            "missing tsconfig must not produce MalformedTsconfig: {diagnostics:?}"
        );
    }

    #[test]
    fn shallow_scan_malformed_package_json_stays_silent() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let scratch = dir.path().join("scratch");
        std::fs::create_dir_all(&scratch).unwrap();
        std::fs::write(scratch.join("package.json"), r"{not valid json}").unwrap();

        let result =
            discover_workspaces_with_diagnostics(dir.path(), &crate::IgnorePatternSet::empty());
        let (_, diagnostics) = result.expect("no root package.json is OK");

        assert!(
            !diagnostics
                .iter()
                .any(|d| matches!(d.kind, WorkspaceDiagnosticKind::MalformedPackageJson { .. })),
            "shallow-scan malformed must stay silent: {diagnostics:?}"
        );
    }

    #[test]
    fn mixed_valid_and_malformed_workspaces_partial_recovery() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let pkg_good = dir.path().join("packages").join("good");
        let pkg_bad = dir.path().join("packages").join("bad");
        std::fs::create_dir_all(&pkg_good).unwrap();
        std::fs::create_dir_all(&pkg_bad).unwrap();
        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/*"]}"#,
        )
        .unwrap();
        std::fs::write(pkg_good.join("package.json"), r#"{"name": "good"}"#).unwrap();
        std::fs::write(pkg_bad.join("package.json"), r"{,").unwrap();

        let result =
            discover_workspaces_with_diagnostics(dir.path(), &crate::IgnorePatternSet::empty());
        let (workspaces, diagnostics) = result.expect("root package.json is valid");

        assert_eq!(workspaces.len(), 1);
        assert_eq!(workspaces[0].name, "good");
        assert_eq!(diagnostics.len(), 1);
        assert_eq!(diagnostics[0].kind.id(), "malformed-package-json");
    }

    #[test]
    fn discover_workspaces_back_compat_drops_diagnostics_and_errors() {
        let dir = tempfile::tempdir().expect("create temp dir");
        std::fs::write(dir.path().join("package.json"), r"{bad json").unwrap();

        let workspaces = discover_workspaces(dir.path());
        assert!(
            workspaces.is_empty(),
            "back-compat wrapper returns empty on root-malformed: {workspaces:?}"
        );
    }

    #[test]
    fn discovers_deno_workspace_members_without_package_json() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let core = dir.path().join("packages").join("core");
        let app = dir.path().join("apps").join("desktop");
        std::fs::create_dir_all(&core).unwrap();
        std::fs::create_dir_all(&app).unwrap();

        std::fs::write(
            dir.path().join("deno.json"),
            r#"{
              "workspace": ["./apps/*", "./packages/*"],
              "imports": { "@std/assert": "jsr:@std/assert@1" }
            }"#,
        )
        .unwrap();
        std::fs::write(
            core.join("deno.json"),
            r#"{
              "name": "@fallow/core",
              "exports": { ".": "./mod.ts", "./result": "./result.ts" }
            }"#,
        )
        .unwrap();
        std::fs::write(
            app.join("deno.json"),
            r#"{ "name": "@fallow/desktop", "exports": { ".": "./main.ts" } }"#,
        )
        .unwrap();

        let workspaces = discover_workspaces(dir.path());
        let mut names: Vec<_> = workspaces.iter().map(|w| w.name.as_str()).collect();
        names.sort_unstable();
        assert_eq!(names, vec!["@fallow/core", "@fallow/desktop"]);
        assert!(
            workspaces
                .iter()
                .all(|w| !w.root.join("package.json").exists())
        );
        assert!(
            workspaces.iter().all(|w| !w.is_internal_dependency),
            "Deno packages need dependency evidence before inventory marks them internal"
        );
    }

    #[test]
    fn malformed_root_deno_config_returns_load_error() {
        let dir = tempfile::tempdir().expect("create temp dir");
        std::fs::write(dir.path().join("deno.jsonc"), "{ workspace: [ }").unwrap();

        let result =
            discover_workspaces_with_diagnostics(dir.path(), &crate::IgnorePatternSet::empty());

        match result {
            Err(WorkspaceLoadError::MalformedRootDenoConfig { path, error }) => {
                assert!(path.ends_with("deno.jsonc"));
                assert!(!error.is_empty(), "underlying parse error is preserved");
            }
            other => panic!("expected MalformedRootDenoConfig, got {other:?}"),
        }
    }

    #[test]
    fn deno_and_npm_workspace_patterns_are_additive() {
        let dir = tempfile::tempdir().expect("create temp dir");
        let a = dir.path().join("packages").join("a");
        let b = dir.path().join("packages").join("b");
        std::fs::create_dir_all(&a).unwrap();
        std::fs::create_dir_all(&b).unwrap();

        std::fs::write(
            dir.path().join("package.json"),
            r#"{"workspaces": ["packages/a"]}"#,
        )
        .unwrap();
        std::fs::write(a.join("package.json"), r#"{"name": "a"}"#).unwrap();
        std::fs::write(
            dir.path().join("deno.json"),
            r#"{"workspace": ["./packages/b"]}"#,
        )
        .unwrap();
        std::fs::write(b.join("deno.json"), r#"{"name": "b"}"#).unwrap();

        let workspaces = discover_workspaces(dir.path());
        let mut names: Vec<_> = workspaces.iter().map(|w| w.name.as_str()).collect();
        names.sort_unstable();
        assert_eq!(names, vec!["a", "b"]);
    }

    /// Issue #2366: `package.json` `workspaces` and `pnpm-workspace.yaml`
    /// `packages` are additive sources, so one glob declared in both is walked
    /// twice. Deduplicating at the discovery choke point is what lets the JSON
    /// envelopes report one entry per distinct matching pattern and lets the
    /// aggregated stderr warning count the directories it actually found.
    #[test]
    fn one_glob_declared_in_two_manifests_reports_each_directory_once() {
        let dir = tempfile::tempdir().expect("create temp dir");
        write_two_manifest_glob_project(dir.path());

        let (_, diagnostics) =
            discover_workspaces_with_diagnostics(dir.path(), &crate::IgnorePatternSet::empty())
                .expect("root package.json is valid");

        let mut reported: Vec<(String, String)> = diagnostics
            .iter()
            .filter_map(|diagnostic| match &diagnostic.kind {
                WorkspaceDiagnosticKind::GlobMatchedNoPackageJson { pattern } => Some((
                    pattern.clone(),
                    diagnostic
                        .path
                        .strip_prefix(dir.path())
                        .unwrap_or(&diagnostic.path)
                        .display()
                        .to_string()
                        .replace('\\', "/"),
                )),
                _ => None,
            })
            .collect();
        reported.sort();

        assert_eq!(
            reported,
            vec![
                ("pkgs/*".to_owned(), "pkgs/aaa".to_owned()),
                ("pkgs/*".to_owned(), "pkgs/bbb".to_owned()),
            ],
            "one glob spelled two ways is one diagnostic per directory: {diagnostics:?}"
        );
    }

    /// The control for the fold above: the key is the WHOLE kind, payload
    /// included. Two overlapping globs each report the same package-less
    /// directory with their own `pattern`, and both survive.
    #[test]
    fn overlapping_globs_still_report_one_directory_once_per_pattern() {
        let dir = tempfile::tempdir().expect("create temp dir");
        std::fs::create_dir_all(dir.path().join("pkgs/aaa")).unwrap();
        std::fs::write(
            dir.path().join("package.json"),
            r#"{"name":"overlap-root","private":true,"workspaces":["pkgs/*","pkgs/a*"]}"#,
        )
        .unwrap();
        std::fs::write(dir.path().join("pkgs/aaa/readme.txt"), "no package.json\n").unwrap();

        let (_, diagnostics) =
            discover_workspaces_with_diagnostics(dir.path(), &crate::IgnorePatternSet::empty())
                .expect("root package.json is valid");

        let patterns: Vec<&str> = diagnostics
            .iter()
            .filter_map(|diagnostic| match &diagnostic.kind {
                WorkspaceDiagnosticKind::GlobMatchedNoPackageJson { pattern } => {
                    Some(pattern.as_str())
                }
                _ => None,
            })
            .collect();

        assert_eq!(
            patterns,
            ["pkgs/*", "pkgs/a*"],
            "distinct patterns are distinct diagnostics: {diagnostics:?}"
        );
    }
}