claudix 0.1.4

mod input;
mod install;
mod watch;

use std::collections::{HashMap, HashSet};
use std::fs;
use std::io::{self, Write};
use std::path::{Path, PathBuf};
use std::sync::Arc;

use serde::Serialize;

use crate::config;
use crate::error::{ClaudixError, RecoveryHint, Result};
use crate::hooks::HookEvent;
use crate::search::SearchQuery;
use crate::search::duplicates::{self, LabeledChunk};
pub use crate::search::duplicates::{DuplicateChunk, DuplicatePair};
use crate::store::{IndexLockGuard, Store};
use crate::types::{RelativePath, path_prefix_matches};
use crate::{Claudix, IndexFileStatus, IndexProgress};

use input::{
    parse_language_filter, parse_path_prefix, validate_search_query, validate_search_top_k,
};

/// Maximum number of top identifiers surfaced per directory in `OverviewOutput`.
const TOP_IDENTIFIERS_CAP: usize = 8;

/// Default cosine-similarity floor for [`run_find_duplicates`]. Higher = stricter / fewer pairs.
pub(crate) const DEFAULT_MIN_SIMILARITY: f32 = 0.85;
/// Default maximum number of duplicate pairs returned by [`run_find_duplicates`].
pub(crate) const DEFAULT_DUPLICATE_LIMIT: usize = 50;

pub use install::{run_install, setup_state};
pub use watch::run_watch;

#[derive(Debug, Clone, PartialEq, Serialize)]
pub struct SearchHit {
    /// Canonical path of the repo this hit came from.
    pub repo: String,
    pub file_path: String,
    pub language: String,
    pub kind: String,
    pub name: Option<String>,
    pub line_start: u32,
    pub line_end: u32,
    pub score: f32,
    pub stale: bool,
    pub snippet: String,
}

#[derive(Debug, Clone, PartialEq, Serialize)]
pub struct DirectoryGroup {
    /// Canonical repo path the directory belongs to. Same-named directories in
    /// different repos do not merge — the group key is `(repo, directory)`.
    pub repo: String,
    pub directory: String,
    pub hits: Vec<SearchHit>,
}

#[derive(Debug, Clone, PartialEq, Serialize)]
pub struct SearchOutput {
    pub groups: Vec<DirectoryGroup>,
    /// Repos that could not be searched (unindexed, mismatched, missing).
    /// Empty for single-repo searches.
    pub repo_errors: Vec<RepoError>,
}

#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
pub struct IndexOutput {
    pub file_count: usize,
    pub chunk_count: usize,
}

pub struct StderrIndexProgress;

impl IndexProgress for StderrIndexProgress {
    fn file(&mut self, path: &RelativePath, status: IndexFileStatus) -> Result<()> {
        let mut stderr = io::stderr().lock();
        match status {
            IndexFileStatus::Indexed => writeln!(stderr, "indexed {}", path.as_str())?,
            IndexFileStatus::Verified => writeln!(stderr, "verified {}", path.as_str())?,
            IndexFileStatus::Skipped(reason) => {
                writeln!(stderr, "skipped {}: {reason}", path.as_str())?
            }
        }
        stderr.flush()?;
        Ok(())
    }
}

struct FileIndexProgress {
    writer: io::BufWriter<fs::File>,
}

impl FileIndexProgress {
    fn try_open(log_dir: &Path) -> Option<Self> {
        fs::create_dir_all(log_dir).ok()?;
        fs::File::create(log_dir.join("index.log"))
            .ok()
            .map(|f| Self {
                writer: io::BufWriter::new(f),
            })
    }
}

impl IndexProgress for FileIndexProgress {
    fn file(&mut self, path: &RelativePath, status: IndexFileStatus) -> Result<()> {
        match status {
            IndexFileStatus::Indexed => writeln!(self.writer, "indexed {}", path.as_str())?,
            IndexFileStatus::Verified => writeln!(self.writer, "verified {}", path.as_str())?,
            IndexFileStatus::Skipped(reason) => {
                writeln!(self.writer, "skipped {}: {reason}", path.as_str())?
            }
        }
        self.writer.flush()?;
        Ok(())
    }
}

#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
pub struct ClearOutput {
    pub cleared: bool,
}

#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
pub struct StatusOutput {
    pub chunk_count: usize,
    pub file_count: usize,
    pub model: Option<String>,
    pub dimensions: Option<u16>,
    pub last_full_index_at: Option<String>,
    pub last_incremental_at: Option<String>,
    /// True when the index is missing or older than `reindex_after_hours`.
    pub stale: bool,
}

#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
pub struct DoctorOutput {
    pub project_root: String,
    pub index_present: bool,
    pub chunk_count: usize,
    pub file_count: usize,
    pub model: Option<String>,
    pub dimensions: Option<u16>,
    pub embedding_provider: String,
    pub embedding_healthy: bool,
    /// Why the embedding check failed, when `embedding_healthy` is false and the
    /// cause is not a model mismatch — auth, timeout, unreachable, HTTP status.
    pub embedding_error: Option<String>,
    /// True when the stored index model differs from the active config model.
    /// Distinct from `embedding_healthy = false` caused by the server being unreachable.
    pub embedding_model_mismatch: bool,
    /// Dev-only flag from config: when on, `scripts/ensure-binary.sh` runs the
    /// `cargo install` binary instead of the downloaded release.
    pub development_mode: bool,
    /// Absolute path of the binary serving this command, so it is unambiguous
    /// which build is running (cargo vs. cached release) during development.
    pub binary_path: String,
}

#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
pub struct InstallOutput {
    pub plugin_root: String,
    pub binary_path: String,
    pub config_path: String,
    pub wrote_config: bool,
    pub embedding_healthy: bool,
}

#[derive(Debug, Clone, PartialEq, Eq)]
pub enum SetupState {
    Ready,
    Missing(Vec<&'static str>),
}

/// Per-language chunk count within a directory.
#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
pub struct LanguageCount {
    pub language: String,
    pub chunk_count: usize,
}

/// Aggregated stats for one immediate parent directory in the index.
#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
pub struct DirectoryRollup {
    /// Repo-relative directory path, or `"."` for root-level files.
    pub path: String,
    pub file_count: usize,
    pub chunk_count: usize,
    /// Languages present in this directory, sorted by chunk count desc then name asc.
    pub languages: Vec<LanguageCount>,
    /// Most frequent non-empty chunk names, capped at [`TOP_IDENTIFIERS_CAP`], sorted by
    /// frequency desc then name asc.
    pub top_identifiers: Vec<String>,
}

/// Structural map of the indexed repo, grouped by immediate parent directory.
#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
pub struct OverviewOutput {
    /// One entry per directory, sorted by path ascending.
    pub directories: Vec<DirectoryRollup>,
    /// Distinct files in the filtered view.
    pub file_count: usize,
    /// Total chunks in the filtered view.
    pub chunk_count: usize,
}

/// A repo that could not contribute to the duplicate scan, with a reason.
#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
pub struct RepoError {
    pub repo: String,
    pub error: String,
}

/// Result of [`run_find_duplicates`].
///
/// Partial success is intentional: indexed repos produce pairs even when
/// some listed repos errored.
#[derive(Debug, Clone, PartialEq, Serialize)]
pub struct DuplicatesOutput {
    pub pairs: Vec<DuplicatePair>,
    pub repo_errors: Vec<RepoError>,
}

pub async fn run_overview(
    project_root: impl AsRef<Path>,
    path_prefix: Option<String>,
) -> Result<OverviewOutput> {
    let project_root = canonical_project_root(project_root.as_ref())?;
    let config = config::load(&project_root)?;
    let store = Store::new(&project_root, &config)?;

    // Validate and normalize the path prefix the same way search does.
    let prefix: Option<RelativePath> = parse_path_prefix(path_prefix)?;

    let chunks = store.read_chunks().await?;

    // Group chunks by the immediate parent directory of their file_path.
    // Splitting on `/` is sound because file_path comes from RelativePath,
    // which forward-slash-normalizes on construction on every platform.
    // Aggregation is light counting over an already-loaded Vec — no spawn_blocking needed.
    let mut dir_files: HashMap<String, std::collections::HashSet<String>> = HashMap::new();
    let mut dir_chunks: HashMap<String, usize> = HashMap::new();
    let mut dir_languages: HashMap<String, HashMap<String, usize>> = HashMap::new();
    let mut dir_names: HashMap<String, HashMap<String, usize>> = HashMap::new();

    for chunk in &chunks {
        // Apply path-prefix filter, consistent with `apply_filters` in search.
        if let Some(ref prefix) = prefix
            && !path_prefix_matches(&chunk.file_path, prefix.as_str())
        {
            continue;
        }

        let dir = immediate_parent_dir(&chunk.file_path);
        dir_files
            .entry(dir.clone())
            .or_default()
            .insert(chunk.file_path.clone());
        *dir_chunks.entry(dir.clone()).or_insert(0) += 1;
        *dir_languages
            .entry(dir.clone())
            .or_default()
            .entry(chunk.language.clone())
            .or_insert(0) += 1;
        if let Some(ref name) = chunk.name
            && !name.is_empty()
        {
            *dir_names
                .entry(dir.clone())
                .or_default()
                .entry(name.clone())
                .or_insert(0) += 1;
        }
    }

    let mut directories: Vec<DirectoryRollup> = dir_files
        .keys()
        .map(|dir| {
            let file_count = dir_files[dir].len();
            let chunk_count = dir_chunks[dir];

            let mut languages: Vec<LanguageCount> = dir_languages
                .get(dir)
                .map(|lang_map| {
                    lang_map
                        .iter()
                        .map(|(lang, count)| LanguageCount {
                            language: lang.clone(),
                            chunk_count: *count,
                        })
                        .collect()
                })
                .unwrap_or_default();
            // Sort by chunk count desc, then language name asc for determinism.
            languages.sort_by(|a, b| {
                b.chunk_count
                    .cmp(&a.chunk_count)
                    .then_with(|| a.language.cmp(&b.language))
            });

            let top_identifiers: Vec<String> = dir_names
                .get(dir)
                .map(|name_map| {
                    let mut pairs: Vec<(&String, usize)> =
                        name_map.iter().map(|(n, c)| (n, *c)).collect();
                    // Frequency desc, then name asc for determinism, then cap.
                    pairs.sort_by(|(a_name, a_count), (b_name, b_count)| {
                        b_count.cmp(a_count).then_with(|| a_name.cmp(b_name))
                    });
                    pairs.truncate(TOP_IDENTIFIERS_CAP);
                    pairs.into_iter().map(|(name, _)| name.clone()).collect()
                })
                .unwrap_or_default();

            DirectoryRollup {
                path: dir.clone(),
                file_count,
                chunk_count,
                languages,
                top_identifiers,
            }
        })
        .collect();

    // Sort directories by path ascending for deterministic, navigable output.
    directories.sort_by(|a, b| a.path.cmp(&b.path));

    let file_count: usize = directories.iter().map(|d| d.file_count).sum();
    let chunk_count: usize = directories.iter().map(|d| d.chunk_count).sum();

    Ok(OverviewOutput {
        directories,
        file_count,
        chunk_count,
    })
}

/// Read the embedding identity (model, dimensions) from a repo's manifest without
/// validating against any reference. Used to bootstrap the reference for the first
/// repo in a multi-repo scan.
async fn peek_manifest_identity(repo_path: &str) -> std::result::Result<(String, u16), RepoError> {
    let config = config::load(std::path::Path::new(repo_path)).map_err(|e| RepoError {
        repo: repo_path.to_owned(),
        error: e.to_string(),
    })?;

    let store = Store::new(repo_path, &config).map_err(|e| RepoError {
        repo: repo_path.to_owned(),
        error: e.to_string(),
    })?;

    let manifest = store.read_manifest().map_err(|e| RepoError {
        repo: store.project_root().display().to_string(),
        error: e.to_string(),
    })?;

    match manifest {
        Some(m) if m.chunk_count > 0 => Ok((m.embedding_model, m.dimensions)),
        _ => Err(RepoError {
            repo: repo_path.to_owned(),
            error: "not indexed".to_owned(),
        }),
    }
}

/// Open a repo read-only, confirm it is indexed and dimension-compatible,
/// and return its chunks labeled with the canonical repo path.
///
/// This helper is intentionally read-only: it calls only `Store::new`,
/// `read_manifest`, and `read_chunks` — never `ensure_layout` or `write_manifest`.
/// Feature 6 (cross-repo search) reuses this to load remote repos without writing.
///
/// Returns `Err(RepoError)` when the repo path is invalid, the index is absent,
/// the chunk count is zero, or the embedding identity (model + dimensions) does
/// not match `ref_model`/`ref_dims`.
pub(crate) async fn load_repo_chunks_readonly(
    repo_path: &str,
    ref_model: &str,
    ref_dims: u16,
) -> std::result::Result<(String, Vec<crate::store::StoredChunk>), RepoError> {
    let config = config::load(std::path::Path::new(repo_path)).map_err(|e| RepoError {
        repo: repo_path.to_owned(),
        error: e.to_string(),
    })?;

    let store = Store::new(repo_path, &config).map_err(|e| RepoError {
        repo: repo_path.to_owned(),
        error: e.to_string(),
    })?;

    // Use the canonical path the store resolved to as the stable repo key.
    let canonical_repo = store.project_root().display().to_string();

    let manifest = store
        .validate_manifest_compatibility(ref_model, ref_dims)
        .map_err(|e| RepoError {
            repo: canonical_repo.clone(),
            error: e.to_string(),
        })?;

    let Some(manifest) = manifest else {
        return Err(RepoError {
            repo: canonical_repo,
            error: "not indexed".to_owned(),
        });
    };

    if manifest.chunk_count == 0 {
        return Err(RepoError {
            repo: canonical_repo,
            error: "not indexed".to_owned(),
        });
    }

    let chunks = store.read_chunks().await.map_err(|e| RepoError {
        repo: canonical_repo.clone(),
        error: e.to_string(),
    })?;

    if chunks.is_empty() {
        return Err(RepoError {
            repo: canonical_repo,
            error: "index chunks missing".to_owned(),
        });
    }

    Ok((canonical_repo, chunks))
}

/// Find near-duplicate code chunks within the active repo or across an explicit list of repos.
///
/// When `repos` is `None` or empty, only `project_root` is scanned.
/// When `repos` is non-empty, EXACTLY those paths are used — `project_root` is NOT
/// auto-added; the caller decides what to include.
///
/// The reference embedding identity (model + dimensions) is established from the
/// manifest of the first repo that loads successfully. All subsequent repos must
/// match that identity or they become `RepoError`s.
///
/// The O(n²) detection scan runs inside `tokio::task::spawn_blocking`.
pub async fn run_find_duplicates(
    project_root: impl AsRef<Path>,
    min_similarity: Option<f32>,
    limit: Option<usize>,
    repos: Option<Vec<String>>,
) -> Result<DuplicatesOutput> {
    let project_root = canonical_project_root(project_root.as_ref())?;
    let min_similarity = min_similarity.unwrap_or(DEFAULT_MIN_SIMILARITY);
    if !min_similarity.is_finite() || !(0.0..=1.0).contains(&min_similarity) {
        return Err(ClaudixError::ConfigInvalid {
            message: "min_similarity must be between 0 and 1".to_owned(),
            recovery: RecoveryHint("Use a finite min_similarity between 0 and 1"),
        });
    }
    let limit = limit.unwrap_or(DEFAULT_DUPLICATE_LIMIT);
    if limit == 0 {
        return Err(ClaudixError::ConfigInvalid {
            message: "limit must be at least 1".to_owned(),
            recovery: RecoveryHint("Use a positive limit"),
        });
    }

    // Determine which repo paths to scan.
    let repo_paths: Vec<String> = match repos {
        Some(list) if !list.is_empty() => list,
        _ => vec![project_root.display().to_string()],
    };

    let mut all_chunks: Vec<crate::store::StoredChunk> = Vec::new();
    let mut repo_labels: Vec<String> = Vec::new();
    let mut repo_errors: Vec<RepoError> = Vec::new();

    // The reference embedding identity is taken from the first repo whose manifest
    // loads successfully. Subsequent repos must match or they become RepoErrors.
    let mut ref_identity: Option<(String, u16)> = None;

    for path in &repo_paths {
        // Resolve reference identity lazily from the first successful manifest.
        if ref_identity.is_none() {
            match peek_manifest_identity(path).await {
                Ok(identity) => ref_identity = Some(identity),
                Err(err) => {
                    repo_errors.push(err);
                    continue;
                }
            }
        }

        let Some((ref_model, ref_dims)) = ref_identity.as_ref() else {
            continue;
        };
        match load_repo_chunks_readonly(path, ref_model, *ref_dims).await {
            Ok((canonical, chunks)) => {
                for _ in &chunks {
                    repo_labels.push(canonical.clone());
                }
                all_chunks.extend(chunks);
            }
            Err(err) => {
                repo_errors.push(err);
            }
        }
    }

    if all_chunks.is_empty() {
        return Ok(DuplicatesOutput {
            pairs: Vec::new(),
            repo_errors,
        });
    }

    // Build the labeled slice for the detection scan.
    // `spawn_blocking` keeps the O(n²) work off the async executor.
    let pairs = tokio::task::spawn_blocking(move || {
        let labeled: Vec<LabeledChunk<'_>> = all_chunks
            .iter()
            .zip(repo_labels.iter())
            .map(|(chunk, repo)| LabeledChunk {
                repo: repo.as_str(),
                chunk,
            })
            .collect();
        duplicates::find_duplicates(&labeled, min_similarity, limit)
    })
    .await
    .map_err(|e| ClaudixError::Store(format!("duplicate scan task failed: {e}")))?;

    Ok(DuplicatesOutput { pairs, repo_errors })
}

/// Returns the immediate parent directory of a repo-relative file path,
/// normalized to forward slashes. Root-level files (no `/`) return `"."`.
///
/// Used by `run_overview` for directory grouping. Feature 5 can reuse this
/// for grouping search results.
pub(crate) fn immediate_parent_dir(file_path: &str) -> String {
    match file_path.rfind('/') {
        Some(pos) => file_path[..pos].to_owned(),
        None => ".".to_owned(),
    }
}

pub async fn run_search(
    project_root: impl AsRef<Path>,
    query: String,
    top_k: Option<usize>,
    language_filter: Option<Vec<String>>,
    path_prefix: Option<String>,
    repos: Option<Vec<String>>,
) -> Result<SearchOutput> {
    validate_search_query(&query)?;
    let project_root = canonical_project_root(project_root.as_ref())?;
    let config = config::load(&project_root)?;
    let top_k = top_k.unwrap_or(config.search.top_k);
    validate_search_top_k(top_k)?;
    // Active project is always in scope; union the config cross_repos with the
    // per-call repos, deduped at search time by canonical path.
    let repos = effective_cross_repos(&config.search.cross_repos, repos);
    let claudix = Claudix::new(project_root, Arc::new(config)).await?;

    run_search_with_claudix(&claudix, query, top_k, language_filter, path_prefix, repos).await
}

/// Union the configured `cross_repos` with the per-call `repos`, preserving
/// order and dropping exact-string duplicates. Canonical-path dedup happens
/// later in the searcher (it needs to resolve each path through the store).
fn effective_cross_repos(cross_repos: &[String], repos: Option<Vec<String>>) -> Vec<String> {
    let mut seen = HashSet::new();
    cross_repos
        .iter()
        .cloned()
        .chain(repos.into_iter().flatten())
        .filter(|repo| seen.insert(repo.clone()))
        .collect()
}

pub async fn run_index(project_root: impl AsRef<Path>, progress: bool) -> Result<IndexOutput> {
    let session = IndexSession::new(project_root).await?;
    let log_dir = session
        .claudix
        .project_root()
        .join(&session.claudix.config().paths.log_dir);
    let mut stderr_progress = StderrIndexProgress;
    let mut file_progress = (!progress)
        .then(|| FileIndexProgress::try_open(&log_dir))
        .flatten();
    let progress: &mut dyn IndexProgress = if progress {
        &mut stderr_progress
    } else if let Some(ref mut fp) = file_progress {
        fp
    } else {
        &mut ()
    };
    let stats = session.claudix.index_full(progress).await?;

    Ok(IndexOutput {
        file_count: stats.file_count,
        chunk_count: stats.chunk_count,
    })
}

struct IndexSession {
    claudix: Claudix,
    _lock: IndexLockGuard,
}

impl IndexSession {
    async fn new(project_root: impl AsRef<Path>) -> Result<Self> {
        let project_root = canonical_project_root(project_root.as_ref())?;
        require_git_repo(&project_root)?;
        let config = config::load(&project_root)?;
        let store = Store::new(&project_root, &config)?;
        let lock = store
            .acquire_index_lock()
            .ok_or_else(|| crate::error::ClaudixError::Store("index already running".to_owned()))?;
        let claudix = match Claudix::new(project_root.clone(), Arc::new(config.clone())).await {
            Ok(claudix) => claudix,
            Err(error) if requires_clean_reindex(&error) => {
                store.clear_chunks(&config).await?;
                Claudix::new(project_root, Arc::new(config)).await?
            }
            Err(error) => return Err(error),
        };

        Ok(Self {
            claudix,
            _lock: lock,
        })
    }
}

pub async fn run_status(project_root: impl AsRef<Path>) -> Result<StatusOutput> {
    let project_root = canonical_project_root(project_root.as_ref())?;
    let config = config::load(&project_root)?;
    let store = Store::new(&project_root, &config)?;
    status_from_store(&store, &config).await
}

pub async fn run_reindex_file(
    project_root: impl AsRef<Path>,
    path: impl AsRef<Path>,
) -> Result<IndexOutput> {
    let project_root = canonical_project_root(project_root.as_ref())?;
    let config = config::load(&project_root)?;
    let store = Store::new(&project_root, &config)?;

    // Block on the shared chunk-writer lock instead of short-circuiting on a
    // running full index: bailing here loses the user's edit until they save
    // again, since the reindex-file child returns 0 with no retry path.
    let _reindex_lock = store.acquire_reindex_lock()?;
    let claudix = Claudix::new(project_root, Arc::new(config)).await?;
    let stats = claudix.reindex_file(path.as_ref()).await?;

    Ok(IndexOutput {
        file_count: stats.file_count,
        chunk_count: stats.chunk_count,
    })
}

pub async fn run_doctor(project_root: impl AsRef<Path>) -> Result<DoctorOutput> {
    let project_root = canonical_project_root(project_root.as_ref())?;
    let config = config::load(&project_root)?;
    let store = Store::new(&project_root, &config)?;
    let status = status_from_store(&store, &config).await?;

    let claudix = Claudix::new(project_root.clone(), Arc::new(config.clone())).await;
    let (embedding_healthy, embedding_model_mismatch, embedding_error) = match claudix {
        Ok(claudix) => match claudix.embedder_health_check().await {
            Ok(()) => (true, false, None),
            Err(error) => (false, false, Some(error.to_string())),
        },
        Err(ClaudixError::EmbeddingModelMismatch { .. }) => (false, true, None),
        Err(error) => (false, false, Some(error.to_string())),
    };

    Ok(DoctorOutput {
        project_root: project_root.display().to_string(),
        index_present: status.chunk_count > 0 || status.model.is_some(),
        chunk_count: status.chunk_count,
        file_count: status.file_count,
        model: status.model,
        dimensions: status.dimensions,
        embedding_provider: match config.embedding.provider {
            config::EmbeddingProvider::Bundled => "bundled".to_owned(),
            config::EmbeddingProvider::Http => "http".to_owned(),
        },
        embedding_healthy,
        embedding_error,
        embedding_model_mismatch,
        development_mode: config.development_mode,
        binary_path: std::env::current_exe()
            .map(|path| path.display().to_string())
            .unwrap_or_else(|_| "<unknown>".to_owned()),
    })
}

pub async fn run_clear_index(project_root: impl AsRef<Path>) -> Result<ClearOutput> {
    let project_root = canonical_project_root(project_root.as_ref())?;
    let config = config::load(&project_root)?;
    let store = Store::new(&project_root, &config)?;
    store.clear_chunks(&config).await?;

    Ok(ClearOutput { cleared: true })
}

fn requires_clean_reindex(error: &ClaudixError) -> bool {
    matches!(
        error,
        ClaudixError::SchemaMismatch { .. }
            | ClaudixError::EmbeddingModelMismatch { .. }
            | ClaudixError::DimensionMismatch { .. }
    )
}

pub fn parse_hook_event(value: &str) -> Result<HookEvent> {
    match value {
        "SessionStart" => Ok(HookEvent::SessionStart),
        "PostToolUse" => Ok(HookEvent::PostToolUse),
        "PreToolUse" => Ok(HookEvent::PreToolUse),
        "UserPromptSubmit" => Ok(HookEvent::UserPromptSubmit),
        _ => Err(ClaudixError::ConfigInvalid {
            message: format!("unknown hook event: {value}"),
            recovery: RecoveryHint(
                "Use one of: SessionStart, PostToolUse, PreToolUse, UserPromptSubmit",
            ),
        }),
    }
}

async fn run_search_with_claudix(
    claudix: &Claudix,
    query: String,
    top_k: usize,
    language_filter: Option<Vec<String>>,
    path_prefix: Option<String>,
    repos: Vec<String>,
) -> Result<SearchOutput> {
    let query = SearchQuery {
        query,
        top_k,
        language_filter: parse_language_filter(language_filter)?,
        path_prefix: parse_path_prefix(path_prefix)?,
        repos,
    };
    let found = claudix.search(query).await?;

    // Walk hits in score order (results is already score-desc from search).
    // Bucket by (repo, directory) preserving first-seen order so the first key
    // encountered owns the top hit — groups naturally ordered by best score.
    // Keying on repo too keeps same-named directories in different repos apart.
    let mut group_index: Vec<(String, String)> = Vec::new();
    let mut grouped: HashMap<(String, String), Vec<SearchHit>> = HashMap::new();

    for result in found.results {
        let dir = immediate_parent_dir(result.chunk.file_path.as_str());
        let key = (result.repo.clone(), dir);
        let hit = SearchHit {
            repo: result.repo,
            file_path: result.chunk.file_path.to_string(),
            language: result.chunk.language.to_string(),
            kind: result.chunk.kind.to_string(),
            name: result.chunk.name,
            line_start: result.chunk.line_range.start,
            line_end: result.chunk.line_range.end,
            score: result.score,
            stale: result.stale,
            snippet: result.chunk.content,
        };
        if !grouped.contains_key(&key) {
            group_index.push(key.clone());
        }
        grouped.entry(key).or_default().push(hit);
    }

    let groups = group_index
        .into_iter()
        .filter_map(|key| {
            let hits = grouped.remove(&key)?;
            let (repo, directory) = key;
            Some(DirectoryGroup {
                repo,
                directory,
                hits,
            })
        })
        .collect();

    Ok(SearchOutput {
        groups,
        repo_errors: found.repo_errors,
    })
}

async fn status_from_store(store: &Store, config: &crate::config::Config) -> Result<StatusOutput> {
    let manifest = store.read_manifest()?;
    let chunk_count = manifest
        .as_ref()
        .map(|m| m.chunk_count as usize)
        .unwrap_or(0);
    let file_count = manifest
        .as_ref()
        .map(|m| m.file_count as usize)
        .unwrap_or(0);

    let stale = manifest
        .as_ref()
        .map(|m| m.is_stale(config))
        .unwrap_or(true);

    Ok(StatusOutput {
        chunk_count,
        file_count,
        model: manifest
            .as_ref()
            .map(|manifest| manifest.embedding_model.clone()),
        dimensions: manifest.as_ref().map(|manifest| manifest.dimensions),
        last_full_index_at: manifest
            .as_ref()
            .and_then(|manifest| manifest.last_full_index_at.clone()),
        last_incremental_at: manifest
            .as_ref()
            .and_then(|manifest| manifest.last_incremental_at.clone()),
        stale,
    })
}

fn require_git_repo(project_root: &Path) -> Result<()> {
    if !crate::enumeration::is_git_repo(project_root) {
        return Err(ClaudixError::NotAGitRepository {
            path: project_root.to_path_buf(),
            recovery: RecoveryHint("Run claudix index from inside a git repository"),
        });
    }
    Ok(())
}

pub(super) fn canonical_project_root(project_root: &Path) -> Result<PathBuf> {
    project_root.canonicalize().map_err(ClaudixError::from)
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::config::Config;
    use crate::embedding::{Provider, StubProvider};
    use crate::store::{Manifest, Store};
    use crate::types::Dimension;

    mod fixture {
        include!(concat!(
            env!("CARGO_MANIFEST_DIR"),
            "/tests/common/fixture.rs"
        ));
    }

    mod test_support {
        use crate as claudix;

        include!(concat!(
            env!("CARGO_MANIFEST_DIR"),
            "/tests/common/test_support.rs"
        ));
    }

    use fixture::TestFixture;
    use test_support::{index_fixture, stub_config};

    struct CliHarness {
        _fixture: TestFixture,
        claudix: Claudix,
        store: Store,
    }

    fn test_claudix(project_root: PathBuf, config: Config) -> Result<Claudix> {
        let store = Store::new(&project_root, &config)?;
        let config = Arc::new(config);
        let embedder: Arc<dyn Provider> = Arc::new(StubProvider::with_model_id(
            config.embedding.model.clone(),
            Dimension(config.embedding.dimensions),
        ));

        Ok(Claudix::from_parts(project_root, config, embedder, store))
    }

    async fn cli_harness() -> Result<CliHarness> {
        let fixture = TestFixture::new("small_rust")?;
        let config = stub_config();
        let claudix = test_claudix(fixture.root().to_path_buf(), config.clone())?;
        index_fixture(
            &claudix.store,
            claudix.embedder.as_ref(),
            claudix.project_root(),
            &config,
        )
        .await?;
        let store = Store::new(fixture.root(), &config)?;

        Ok(CliHarness {
            _fixture: fixture,
            claudix,
            store,
        })
    }

    #[test]
    fn parse_hook_event_accepts_known_values() {
        let event = parse_hook_event("SessionStart");
        assert!(matches!(event, Ok(HookEvent::SessionStart)));

        let event = parse_hook_event("PostToolUse");
        assert!(matches!(event, Ok(HookEvent::PostToolUse)));

        let event = parse_hook_event("PreToolUse");
        assert!(matches!(event, Ok(HookEvent::PreToolUse)));

        let event = parse_hook_event("UserPromptSubmit");
        assert!(matches!(event, Ok(HookEvent::UserPromptSubmit)));
    }

    #[test]
    fn parse_hook_event_rejects_unknown_values() {
        let result = parse_hook_event("Unknown");
        assert!(matches!(result, Err(ClaudixError::ConfigInvalid { .. })));
    }

    #[tokio::test]
    async fn run_search_returns_ranked_hits() {
        let harness = cli_harness().await;
        assert!(harness.is_ok());
        let harness = harness.ok().unwrap_or_else(|| unreachable!());

        let output = run_search_with_claudix(
            &harness.claudix,
            "add".to_owned(),
            5,
            None,
            None,
            Vec::new(),
        )
        .await;
        assert!(output.is_ok());
        let output = output.ok().unwrap_or_else(|| unreachable!());

        assert!(!output.groups.is_empty());
        let top_hit = &output.groups[0].hits[0];
        assert_eq!(top_hit.name.as_deref(), Some("add"));
        assert_eq!(top_hit.file_path, "src/math.rs");
    }

    #[tokio::test]
    async fn run_search_applies_filters() {
        let harness = cli_harness().await;
        assert!(harness.is_ok());
        let harness = harness.ok().unwrap_or_else(|| unreachable!());

        let output = run_search_with_claudix(
            &harness.claudix,
            "add".to_owned(),
            5,
            Some(vec!["rust".to_owned()]),
            Some(RelativePath::new("src/math").to_string()),
            Vec::new(),
        )
        .await;
        assert!(output.is_ok());
        let output = output.ok().unwrap_or_else(|| unreachable!());

        assert_eq!(output.groups.len(), 1);
        assert_eq!(output.groups[0].hits.len(), 1);
        assert_eq!(output.groups[0].hits[0].file_path, "src/math.rs");
    }

    #[test]
    fn clean_reindex_required_for_manifest_compatibility_errors() {
        assert!(requires_clean_reindex(&ClaudixError::SchemaMismatch {
            store: 0,
            binary: 1,
            recovery: RecoveryHint("reindex"),
        }));
        assert!(requires_clean_reindex(
            &ClaudixError::EmbeddingModelMismatch {
                store_model: "old".to_owned(),
                active_model: "new".to_owned(),
                recovery: RecoveryHint("reindex"),
            }
        ));
        assert!(requires_clean_reindex(&ClaudixError::DimensionMismatch {
            store_dim: 384,
            model_dim: 768,
            recovery: RecoveryHint("reindex"),
        }));
        assert!(!requires_clean_reindex(&ClaudixError::Store(
            "index already running".to_owned()
        )));
    }

    #[tokio::test]
    async fn run_index_clears_model_mismatch_and_reindexes() {
        let fixture = TestFixture::new("small_rust");
        assert!(fixture.is_ok());
        let fixture = fixture.ok().unwrap_or_else(|| unreachable!());
        let config = stub_config();
        let claude_dir = fixture.root().join(".claude");
        assert!(std::fs::create_dir_all(&claude_dir).is_ok());
        let config_text = toml::to_string(&config);
        assert!(config_text.is_ok());
        assert!(
            std::fs::write(
                claude_dir.join("claudix.toml"),
                config_text.ok().unwrap_or_default(),
            )
            .is_ok()
        );

        let store = Store::new(fixture.root(), &config);
        assert!(store.is_ok());
        let store = store.ok().unwrap_or_else(|| unreachable!());
        let old_manifest = Manifest::new("old-model", config.embedding.dimensions);
        assert!(store.write_manifest(&old_manifest).is_ok());

        let output = run_index(fixture.root(), false).await;
        assert!(output.is_ok());
        let output = output.ok().unwrap_or_else(|| unreachable!());
        assert!(output.chunk_count > 0);

        let manifest = store.read_manifest();
        assert!(manifest.is_ok());
        let manifest = manifest.ok().unwrap_or_else(|| unreachable!());
        let manifest = manifest.unwrap_or_else(|| unreachable!());
        assert_eq!(manifest.embedding_model, config.embedding.model);
        assert_eq!(manifest.dimensions, config.embedding.dimensions);
        assert_eq!(manifest.chunk_count as usize, output.chunk_count);
    }

    #[tokio::test]
    async fn run_status_reports_manifest_and_counts() {
        let harness = cli_harness().await;
        assert!(harness.is_ok());
        let harness = harness.ok().unwrap_or_else(|| unreachable!());

        let config = stub_config();
        let status = status_from_store(&harness.store, &config).await;
        assert!(status.is_ok());
        let status = status.ok().unwrap_or_else(|| unreachable!());

        assert_eq!(status.chunk_count, 3);
        assert_eq!(status.file_count, 2);
        assert_eq!(status.model.as_deref(), Some("stub-v1"));
        assert_eq!(status.dimensions, Some(8));
        assert!(!status.stale, "freshly indexed should not be stale");
    }

    #[tokio::test]
    async fn run_doctor_reports_index_and_embedding_health() {
        let harness = cli_harness().await;
        assert!(harness.is_ok());
        let harness = harness.ok().unwrap_or_else(|| unreachable!());

        let config = stub_config();
        let claude_dir = harness.claudix.project_root().join(".claude");
        assert!(std::fs::create_dir_all(&claude_dir).is_ok());
        let config_text = toml::to_string(&config);
        assert!(config_text.is_ok());
        assert!(
            std::fs::write(
                claude_dir.join("claudix.toml"),
                config_text.ok().unwrap_or_default(),
            )
            .is_ok()
        );

        let output = run_doctor(harness.claudix.project_root()).await;
        assert!(output.is_ok());
        let output = output.ok().unwrap_or_else(|| unreachable!());

        assert!(output.index_present);
        assert_eq!(output.chunk_count, 3);
        assert_eq!(output.file_count, 2);
        assert_eq!(output.model.as_deref(), Some("stub-v1"));
        assert_eq!(output.embedding_provider, "bundled");
        assert!(output.embedding_healthy);
    }

    #[tokio::test]
    async fn run_overview_returns_src_directory_rollup() {
        let harness = cli_harness().await;
        assert!(harness.is_ok());
        let harness = harness.ok().unwrap_or_else(|| unreachable!());

        let output = run_overview(harness.claudix.project_root(), None).await;
        assert!(output.is_ok());
        let output = output.ok().unwrap_or_else(|| unreachable!());

        // The small_rust fixture has files under src/ only.
        let src = output.directories.iter().find(|d| d.path == "src");
        assert!(src.is_some(), "expected a 'src' directory in the rollup");
        let src = src.unwrap_or_else(|| unreachable!());

        // src/ has src/math.rs and src/lib.rs — 2 files, 3 chunks total.
        assert_eq!(src.file_count, 2);
        assert_eq!(src.chunk_count, 3);

        // The fixture is Rust source — rust must appear in languages.
        assert!(
            src.languages.iter().any(|l| l.language == "rust"),
            "expected 'rust' among languages in src/"
        );
    }

    #[tokio::test]
    async fn run_overview_top_identifiers_include_known_fixture_name() {
        let harness = cli_harness().await;
        assert!(harness.is_ok());
        let harness = harness.ok().unwrap_or_else(|| unreachable!());

        let output = run_overview(harness.claudix.project_root(), None).await;
        assert!(output.is_ok());
        let output = output.ok().unwrap_or_else(|| unreachable!());

        let src = output.directories.iter().find(|d| d.path == "src");
        assert!(src.is_some());
        let src = src.unwrap_or_else(|| unreachable!());

        // src/math.rs defines `add` — it must appear in top identifiers.
        assert!(
            src.top_identifiers.iter().any(|name| name == "add"),
            "expected 'add' in top_identifiers for src/; got: {:?}",
            src.top_identifiers,
        );
    }

    #[tokio::test]
    async fn run_overview_path_prefix_narrows_to_subtree() {
        let harness = cli_harness().await;
        assert!(harness.is_ok());
        let harness = harness.ok().unwrap_or_else(|| unreachable!());

        // Prefix "src/math" should match only src/math.rs.
        let output =
            run_overview(harness.claudix.project_root(), Some("src/math".to_owned())).await;
        assert!(output.is_ok());
        let output = output.ok().unwrap_or_else(|| unreachable!());

        // Only chunks from src/math.rs pass the filter; src/lib.rs is excluded.
        assert!(output.file_count < 2, "prefix should exclude src/lib.rs");
        assert!(
            output.chunk_count > 0,
            "at least one chunk from src/math.rs"
        );

        // Every directory in the result must have files that match the prefix.
        for dir in &output.directories {
            assert!(
                path_prefix_matches(&format!("{}/file.rs", dir.path), "src/math")
                    || dir.path == "src",
                "unexpected directory outside prefix: {}",
                dir.path,
            );
        }
    }

    #[tokio::test]
    async fn run_overview_directories_sorted_ascending() {
        let harness = cli_harness().await;
        assert!(harness.is_ok());
        let harness = harness.ok().unwrap_or_else(|| unreachable!());

        let output = run_overview(harness.claudix.project_root(), None).await;
        assert!(output.is_ok());
        let output = output.ok().unwrap_or_else(|| unreachable!());

        let paths: Vec<&str> = output.directories.iter().map(|d| d.path.as_str()).collect();
        let mut sorted = paths.clone();
        sorted.sort();
        assert_eq!(
            paths, sorted,
            "directories must be sorted ascending by path"
        );
    }

    #[test]
    fn immediate_parent_dir_returns_dot_for_root_level_file() {
        assert_eq!(immediate_parent_dir("Cargo.toml"), ".");
        assert_eq!(immediate_parent_dir("lib.rs"), ".");
    }

    #[test]
    fn immediate_parent_dir_extracts_parent_segment() {
        assert_eq!(immediate_parent_dir("src/math.rs"), "src");
        assert_eq!(immediate_parent_dir("src/hooks/mod.rs"), "src/hooks");
    }

    #[tokio::test]
    async fn run_doctor_surfaces_specific_embedding_failure() {
        let fixture = TestFixture::new("small_rust");
        assert!(fixture.is_ok());
        let fixture = fixture.ok().unwrap_or_else(|| unreachable!());

        // Point at a dead port with a non-bundled model so no fallback masks the
        // failure; doctor must report the actual reason, not a flat "unreachable".
        let mut config = stub_config();
        config.embedding.provider = config::EmbeddingProvider::Http;
        config.embedding.endpoint = "http://127.0.0.1:1".to_owned();
        config.embedding.model = "no-fallback-model".to_owned();
        config.embedding.timeout_ms = 100;

        let claude_dir = fixture.root().join(".claude");
        assert!(std::fs::create_dir_all(&claude_dir).is_ok());
        let config_text = toml::to_string(&config);
        assert!(config_text.is_ok());
        assert!(
            std::fs::write(
                claude_dir.join("claudix.toml"),
                config_text.ok().unwrap_or_default(),
            )
            .is_ok()
        );

        let output = run_doctor(fixture.root()).await;
        assert!(output.is_ok());
        let output = output.ok().unwrap_or_else(|| unreachable!());

        assert!(!output.embedding_healthy);
        assert!(!output.embedding_model_mismatch);
        assert!(
            output
                .embedding_error
                .is_some_and(|reason| reason.contains("http://127.0.0.1:1"))
        );
    }

    #[tokio::test]
    async fn search_groups_hits_by_directory_ordered_by_best_score() {
        let harness = cli_harness().await;
        assert!(harness.is_ok());
        let harness = harness.ok().unwrap_or_else(|| unreachable!());

        // "greet add" spans both src/lib.rs and src/math.rs — two directories
        // are both under "src", so we expect exactly one group named "src".
        // The small_rust fixture has all files under src/, so one group.
        let output = run_search_with_claudix(
            &harness.claudix,
            "add greet".to_owned(),
            10,
            None,
            None,
            Vec::new(),
        )
        .await;
        assert!(output.is_ok());
        let output = output.ok().unwrap_or_else(|| unreachable!());

        assert!(!output.groups.is_empty(), "expected at least one group");

        // Every group's directory must equal immediate_parent_dir of its hits.
        for group in &output.groups {
            for hit in &group.hits {
                assert_eq!(
                    immediate_parent_dir(&hit.file_path),
                    group.directory,
                    "hit {} belongs in wrong group",
                    hit.file_path,
                );
            }
        }

        // Hits within each group must be in score-descending order.
        for group in &output.groups {
            let scores: Vec<f32> = group.hits.iter().map(|h| h.score).collect();
            let mut sorted = scores.clone();
            sorted.sort_by(|a, b| b.partial_cmp(a).unwrap_or(std::cmp::Ordering::Equal));
            assert_eq!(
                scores, sorted,
                "hits in group '{}' not score-desc",
                group.directory
            );
        }
    }

    #[tokio::test]
    async fn search_grouping_preserves_top_hit_ranking() {
        let harness = cli_harness().await;
        assert!(harness.is_ok());
        let harness = harness.ok().unwrap_or_else(|| unreachable!());

        let output = run_search_with_claudix(
            &harness.claudix,
            "add".to_owned(),
            5,
            None,
            None,
            Vec::new(),
        )
        .await;
        assert!(output.is_ok());
        let output = output.ok().unwrap_or_else(|| unreachable!());

        assert!(!output.groups.is_empty());

        // The globally top-ranked hit must be groups[0].hits[0] — grouping must
        // not reorder the flat ranking, only reshape its presentation.
        let top = &output.groups[0].hits[0];
        let all_scores: Vec<f32> = output
            .groups
            .iter()
            .flat_map(|g| g.hits.iter().map(|h| h.score))
            .collect();
        let global_max = all_scores.iter().cloned().fold(f32::NEG_INFINITY, f32::max);
        assert_eq!(
            top.score, global_max,
            "top hit in groups[0] must have the globally highest score"
        );
    }

    // ── find_duplicates tests ────────────────────────────────────────────────

    /// Build a temporary fixture with two Rust source files having identical content,
    /// index them with the stub provider, and return (fixture, store) so tests can
    /// call `run_find_duplicates`.
    async fn dup_harness() -> Result<(TestFixture, Store)> {
        let fixture = TestFixture::new("small_rust")?;
        // Write a second file whose content is byte-identical to src/math.rs so
        // the StubProvider (content-hash seed) emits the same vector → cosine = 1.0.
        let dup_content = std::fs::read_to_string(fixture.root().join("src").join("math.rs"))
            .map_err(ClaudixError::from)?;
        std::fs::write(fixture.root().join("src").join("math_copy.rs"), dup_content)
            .map_err(ClaudixError::from)?;

        let config = stub_config();
        let claudix = test_claudix(fixture.root().to_path_buf(), config.clone())?;
        index_fixture(
            &claudix.store,
            claudix.embedder.as_ref(),
            claudix.project_root(),
            &config,
        )
        .await?;
        let store = Store::new(fixture.root(), &config)?;
        Ok((fixture, store))
    }

    #[tokio::test]
    async fn find_duplicates_returns_pair_for_identical_content() {
        let result = dup_harness().await;
        assert!(result.is_ok());
        let (fixture, _store) = result.ok().unwrap_or_else(|| unreachable!());

        let output = run_find_duplicates(fixture.root(), None, None, None).await;
        assert!(output.is_ok());
        let output = output.ok().unwrap_or_else(|| unreachable!());

        // math.rs and math_copy.rs share identical content → identical vectors → cosine 1.0.
        assert!(
            !output.pairs.is_empty(),
            "expected at least one pair from identical file content"
        );
        // Both chunks name the right files.
        let pair = &output.pairs[0];
        let paths = [pair.a.file_path.as_str(), pair.b.file_path.as_str()];
        assert!(
            paths.iter().any(|p| p.contains("math.rs")),
            "expected math.rs in the pair; got {:?}",
            paths
        );
        assert!(
            paths.iter().any(|p| p.contains("math_copy.rs")),
            "expected math_copy.rs in the pair; got {:?}",
            paths
        );
        assert!(
            pair.similarity > 0.99,
            "similarity should be ~1.0 for identical content"
        );
    }

    #[tokio::test]
    async fn find_duplicates_returns_empty_for_unique_repo() {
        // small_rust has lib.rs and math.rs with distinct content → no duplicates.
        let harness = cli_harness().await;
        assert!(harness.is_ok());
        let harness = harness.ok().unwrap_or_else(|| unreachable!());

        let output = run_find_duplicates(harness.claudix.project_root(), None, None, None).await;
        assert!(output.is_ok());
        let output = output.ok().unwrap_or_else(|| unreachable!());

        assert!(
            output.pairs.is_empty(),
            "distinct-content repo should have no duplicates"
        );
    }

    #[tokio::test]
    async fn find_duplicates_threshold_sensitivity() {
        let result = dup_harness().await;
        assert!(result.is_ok());
        let (fixture, _store) = result.ok().unwrap_or_else(|| unreachable!());

        // At threshold 0.99 the identical pair still shows.
        let high = run_find_duplicates(fixture.root(), Some(0.99), None, None).await;
        assert!(high.is_ok());
        let high = high.ok().unwrap_or_else(|| unreachable!());
        assert!(
            !high.pairs.is_empty(),
            "threshold 0.99 should still find identical pair"
        );

        let ceiling = run_find_duplicates(fixture.root(), Some(1.01), None, None).await;
        assert!(
            ceiling.is_err(),
            "threshold > 1.0 must be rejected before scanning"
        );
    }

    #[tokio::test]
    async fn find_duplicates_rejects_invalid_threshold_and_limit() {
        let harness = cli_harness().await;
        assert!(harness.is_ok());
        let harness = harness.ok().unwrap_or_else(|| unreachable!());

        let nan =
            run_find_duplicates(harness.claudix.project_root(), Some(f32::NAN), None, None).await;
        assert!(nan.is_err(), "NaN threshold must be rejected");

        let negative =
            run_find_duplicates(harness.claudix.project_root(), Some(-0.1), None, None).await;
        assert!(negative.is_err(), "negative threshold must be rejected");

        let zero_limit =
            run_find_duplicates(harness.claudix.project_root(), None, Some(0), None).await;
        assert!(zero_limit.is_err(), "zero limit must be rejected");
    }

    #[tokio::test]
    async fn find_duplicates_same_file_not_reported() {
        // Even when identical chunks come from the same file, only cross-file pairs
        // should appear. The dup_harness fixture has math.rs AND math_copy.rs (different
        // files), so the pair is expected — but confirm no entry has a.file_path == b.file_path.
        let result = dup_harness().await;
        assert!(result.is_ok());
        let (fixture, _store) = result.ok().unwrap_or_else(|| unreachable!());

        let output = run_find_duplicates(fixture.root(), Some(0.0), None, None).await;
        assert!(output.is_ok());
        let output = output.ok().unwrap_or_else(|| unreachable!());

        for pair in &output.pairs {
            assert!(
                pair.a.file_path != pair.b.file_path || pair.a.repo != pair.b.repo,
                "same-file pair must not be reported: {} == {}",
                pair.a.file_path,
                pair.b.file_path,
            );
        }
    }

    #[tokio::test]
    async fn find_duplicates_partial_success_on_unindexed_path() {
        let result = dup_harness().await;
        assert!(result.is_ok());
        let (fixture, _store) = result.ok().unwrap_or_else(|| unreachable!());

        let indexed = fixture.root().display().to_string();
        let unindexed = "/tmp/nonexistent-claudix-test-repo-12345".to_owned();

        let output = run_find_duplicates(
            fixture.root(),
            None,
            None,
            Some(vec![indexed, unindexed.clone()]),
        )
        .await;
        assert!(output.is_ok());
        let output = output.ok().unwrap_or_else(|| unreachable!());

        // The unindexed path must produce a RepoError.
        assert!(
            output.repo_errors.iter().any(|e| e.repo == unindexed
                || e.error.contains("not indexed")
                || e.error.contains("No such")),
            "expected a repo_error for the unindexed path; got: {:?}",
            output.repo_errors,
        );
        // Pairs from the indexed repo are still returned.
        assert!(
            !output.pairs.is_empty(),
            "indexed repo should still produce pairs despite the error"
        );
    }

    #[tokio::test]
    async fn load_repo_chunks_readonly_rejects_dimension_mismatch() {
        let harness = cli_harness().await;
        assert!(harness.is_ok());
        let harness = harness.ok().unwrap_or_else(|| unreachable!());

        let root = harness.claudix.project_root().display().to_string();
        // Manifest has dimensions=8 (stub_config). Asking with wrong dims triggers mismatch.
        let result = load_repo_chunks_readonly(&root, "stub-v1", 999).await;
        assert!(
            result.is_err(),
            "mismatched dimensions must produce a RepoError"
        );
        let err = result.err().unwrap_or_else(|| unreachable!());
        assert!(
            err.error.contains("mismatch"),
            "error should mention mismatch; got: {}",
            err.error,
        );
    }

    #[tokio::test]
    async fn load_repo_chunks_readonly_rejects_model_mismatch() {
        let harness = cli_harness().await;
        assert!(harness.is_ok());
        let harness = harness.ok().unwrap_or_else(|| unreachable!());

        let root = harness.claudix.project_root().display().to_string();
        let result = load_repo_chunks_readonly(&root, "different-model", 8).await;
        assert!(result.is_err(), "mismatched model must produce a RepoError");
        let err = result.err().unwrap_or_else(|| unreachable!());
        assert!(
            err.error.contains("mismatch"),
            "error should mention mismatch; got: {}",
            err.error,
        );
    }

    #[tokio::test]
    async fn load_repo_chunks_readonly_rejects_missing_chunk_table() {
        let harness = cli_harness().await;
        assert!(harness.is_ok());
        let harness = harness.ok().unwrap_or_else(|| unreachable!());

        let index_dir = harness.claudix.store.state_dir_path().join("index");
        let remove = std::fs::remove_dir_all(&index_dir);
        assert!(remove.is_ok());

        let root = harness.claudix.project_root().display().to_string();
        let result = load_repo_chunks_readonly(&root, "stub-v1", 8).await;
        assert!(
            result.is_err(),
            "manifest claiming chunks without a chunks table must produce a RepoError"
        );
        let err = result.err().unwrap_or_else(|| unreachable!());
        assert!(
            err.error.contains("chunks missing"),
            "error should mention missing chunks; got: {}",
            err.error,
        );
    }

    #[tokio::test]
    async fn find_duplicates_multi_repo_detects_cross_repo_pair() {
        // Build two separate repos with an identical file, index each, then scan both.
        let fixture_a = TestFixture::new("small_rust");
        assert!(fixture_a.is_ok());
        let fixture_a = fixture_a.ok().unwrap_or_else(|| unreachable!());

        let fixture_b = TestFixture::new("small_rust");
        assert!(fixture_b.is_ok());
        let fixture_b = fixture_b.ok().unwrap_or_else(|| unreachable!());

        let config = stub_config();

        // Index repo A.
        let claudix_a = test_claudix(fixture_a.root().to_path_buf(), config.clone());
        assert!(claudix_a.is_ok());
        let claudix_a = claudix_a.ok().unwrap_or_else(|| unreachable!());
        let index_a = index_fixture(
            &claudix_a.store,
            claudix_a.embedder.as_ref(),
            claudix_a.project_root(),
            &config,
        )
        .await;
        assert!(index_a.is_ok());

        // Index repo B.
        let claudix_b = test_claudix(fixture_b.root().to_path_buf(), config.clone());
        assert!(claudix_b.is_ok());
        let claudix_b = claudix_b.ok().unwrap_or_else(|| unreachable!());
        let index_b = index_fixture(
            &claudix_b.store,
            claudix_b.embedder.as_ref(),
            claudix_b.project_root(),
            &config,
        )
        .await;
        assert!(index_b.is_ok());

        let repo_a = fixture_a.root().display().to_string();
        let repo_b = fixture_b.root().display().to_string();

        // Provide both repos explicitly — active project is NOT auto-added.
        let output = run_find_duplicates(
            fixture_a.root(),
            Some(0.99),
            None,
            Some(vec![repo_a.clone(), repo_b.clone()]),
        )
        .await;
        assert!(output.is_ok());
        let output = output.ok().unwrap_or_else(|| unreachable!());

        assert!(
            output.repo_errors.is_empty(),
            "both repos are indexed; no errors expected"
        );

        // Both repos contain the same fixture content → at least one cross-repo pair.
        assert!(
            !output.pairs.is_empty(),
            "identical fixture content across repos must produce cross-repo pairs"
        );

        // At least one pair must have a.repo != b.repo.
        let has_cross_repo = output.pairs.iter().any(|p| p.a.repo != p.b.repo);
        assert!(
            has_cross_repo,
            "at least one pair must span two different repos"
        );
    }

    // ── cross-repo search tests (Feature 6) ─────────────────────────────────

    /// Write `.claude/claudix.toml` into a fixture so `config::load` doesn't
    /// fall through to the user's global config (which on dev machines may
    /// point at a real embedding model that doesn't match the stub index).
    fn write_fixture_config(project_root: &Path, config: &Config) -> Result<()> {
        let claude_dir = project_root.join(".claude");
        fs::create_dir_all(&claude_dir).map_err(ClaudixError::from)?;
        let text = toml::to_string(config)
            .map_err(|e| ClaudixError::Store(format!("serialize stub config: {e}")))?;
        fs::write(claude_dir.join("claudix.toml"), text).map_err(ClaudixError::from)?;
        Ok(())
    }

    /// Set up two repos backed by the small_rust fixture and indexed with the
    /// same stub model so their vectors are comparable. Returns the canonical
    /// path of each. Caller passes one as `project_root`, the other as `repos`.
    async fn dual_repo_harness() -> Result<(TestFixture, TestFixture, String, String)> {
        let fixture_a = TestFixture::new("small_rust")?;
        let fixture_b = TestFixture::new("small_rust")?;
        let config = stub_config();

        let claudix_a = test_claudix(fixture_a.root().to_path_buf(), config.clone())?;
        index_fixture(
            &claudix_a.store,
            claudix_a.embedder.as_ref(),
            claudix_a.project_root(),
            &config,
        )
        .await?;
        write_fixture_config(fixture_a.root(), &config)?;

        let claudix_b = test_claudix(fixture_b.root().to_path_buf(), config.clone())?;
        index_fixture(
            &claudix_b.store,
            claudix_b.embedder.as_ref(),
            claudix_b.project_root(),
            &config,
        )
        .await?;
        write_fixture_config(fixture_b.root(), &config)?;

        let repo_a = fixture_a.root().display().to_string();
        let repo_b = fixture_b.root().display().to_string();
        Ok((fixture_a, fixture_b, repo_a, repo_b))
    }

    #[tokio::test]
    async fn search_spans_active_and_listed_repo_with_correct_labels() {
        let setup = dual_repo_harness().await;
        assert!(setup.is_ok());
        let (fixture_a, _fixture_b, repo_a, repo_b) = setup.ok().unwrap_or_else(|| unreachable!());

        let output = run_search(
            fixture_a.root(),
            "add".to_owned(),
            Some(10),
            None,
            None,
            Some(vec![repo_b.clone()]),
        )
        .await;
        assert!(output.is_ok(), "cross-repo search failed: {output:?}");
        let output = output.ok().unwrap_or_else(|| unreachable!());

        assert!(output.repo_errors.is_empty(), "no errors expected");

        // Hits surface from both repos, each correctly labeled.
        let from_a = output
            .groups
            .iter()
            .flat_map(|g| g.hits.iter())
            .any(|h| h.repo == repo_a);
        let from_b = output
            .groups
            .iter()
            .flat_map(|g| g.hits.iter())
            .any(|h| h.repo == repo_b);
        assert!(
            from_a,
            "expected at least one hit from active repo {repo_a}"
        );
        assert!(from_b, "expected at least one hit from extra repo {repo_b}");

        // Every hit's repo equals its group's repo.
        for group in &output.groups {
            for hit in &group.hits {
                assert_eq!(
                    hit.repo, group.repo,
                    "hit/group repo mismatch in directory {}",
                    group.directory
                );
            }
        }
    }

    #[tokio::test]
    async fn search_partial_success_on_unindexed_repo() {
        let harness = cli_harness().await;
        assert!(harness.is_ok());
        let harness = harness.ok().unwrap_or_else(|| unreachable!());
        let write = write_fixture_config(harness.claudix.project_root(), &stub_config());
        assert!(write.is_ok());

        let unindexed = "/tmp/nonexistent-claudix-cross-repo-search-9999".to_owned();
        let output = run_search(
            harness.claudix.project_root(),
            "add".to_owned(),
            Some(10),
            None,
            None,
            Some(vec![unindexed.clone()]),
        )
        .await;
        assert!(output.is_ok());
        let output = output.ok().unwrap_or_else(|| unreachable!());

        // Unindexed path surfaces as a RepoError; the active repo still returns hits.
        assert!(
            output.repo_errors.iter().any(|e| e.repo == unindexed
                || e.error.contains("not indexed")
                || e.error.contains("No such")),
            "expected RepoError for unindexed path; got: {:?}",
            output.repo_errors,
        );
        assert!(
            !output.groups.is_empty(),
            "active repo must still produce hits despite the error"
        );
    }

    #[tokio::test]
    async fn search_partial_success_on_model_mismatch() {
        let fixture_a = TestFixture::new("small_rust");
        assert!(fixture_a.is_ok());
        let fixture_a = fixture_a.ok().unwrap_or_else(|| unreachable!());
        let fixture_b = TestFixture::new("small_rust");
        assert!(fixture_b.is_ok());
        let fixture_b = fixture_b.ok().unwrap_or_else(|| unreachable!());

        // Active repo uses stub-v1; extra repo uses stub-v2 → mismatch.
        let config_a = stub_config();
        let config_b = {
            let mut c = stub_config();
            c.embedding.model = "stub-v2".to_owned();
            c
        };

        let claudix_a = test_claudix(fixture_a.root().to_path_buf(), config_a.clone());
        assert!(claudix_a.is_ok());
        let claudix_a = claudix_a.ok().unwrap_or_else(|| unreachable!());
        let _ = index_fixture(
            &claudix_a.store,
            claudix_a.embedder.as_ref(),
            claudix_a.project_root(),
            &config_a,
        )
        .await;
        let write_a = write_fixture_config(fixture_a.root(), &config_a);
        assert!(write_a.is_ok());

        let claudix_b = test_claudix(fixture_b.root().to_path_buf(), config_b.clone());
        assert!(claudix_b.is_ok());
        let claudix_b = claudix_b.ok().unwrap_or_else(|| unreachable!());
        let _ = index_fixture(
            &claudix_b.store,
            claudix_b.embedder.as_ref(),
            claudix_b.project_root(),
            &config_b,
        )
        .await;
        let write_b = write_fixture_config(fixture_b.root(), &config_b);
        assert!(write_b.is_ok());

        let repo_b = fixture_b.root().display().to_string();
        let output = run_search(
            fixture_a.root(),
            "add".to_owned(),
            Some(10),
            None,
            None,
            Some(vec![repo_b.clone()]),
        )
        .await;
        assert!(output.is_ok());
        let output = output.ok().unwrap_or_else(|| unreachable!());

        assert!(
            output
                .repo_errors
                .iter()
                .any(|e| e.error.contains("mismatch")),
            "expected mismatch error for extra repo; got: {:?}",
            output.repo_errors,
        );
        // Active repo still returns hits.
        assert!(
            !output.groups.is_empty(),
            "active repo hits must survive a sibling repo's mismatch"
        );
    }

    #[tokio::test]
    async fn search_dedupes_active_when_listed_in_repos() {
        let harness = cli_harness().await;
        assert!(harness.is_ok());
        let harness = harness.ok().unwrap_or_else(|| unreachable!());
        let write = write_fixture_config(harness.claudix.project_root(), &stub_config());
        assert!(write.is_ok());

        // Baseline: no extra repos.
        let baseline = run_search(
            harness.claudix.project_root(),
            "add".to_owned(),
            Some(10),
            None,
            None,
            None,
        )
        .await;
        assert!(baseline.is_ok());
        let baseline = baseline.ok().unwrap_or_else(|| unreachable!());
        let baseline_hits: usize = baseline.groups.iter().map(|g| g.hits.len()).sum();

        // List the active repo path explicitly — must not double-count.
        let active_path = harness.claudix.project_root().display().to_string();
        let echoed = run_search(
            harness.claudix.project_root(),
            "add".to_owned(),
            Some(10),
            None,
            None,
            Some(vec![active_path]),
        )
        .await;
        assert!(echoed.is_ok());
        let echoed = echoed.ok().unwrap_or_else(|| unreachable!());
        let echoed_hits: usize = echoed.groups.iter().map(|g| g.hits.len()).sum();

        assert_eq!(
            baseline_hits, echoed_hits,
            "listing active path must not duplicate hits"
        );
        assert!(echoed.repo_errors.is_empty());
    }

    #[tokio::test]
    async fn search_groups_separate_same_named_dirs_per_repo() {
        let setup = dual_repo_harness().await;
        assert!(setup.is_ok());
        let (fixture_a, _fixture_b, _repo_a, repo_b) = setup.ok().unwrap_or_else(|| unreachable!());

        let output = run_search(
            fixture_a.root(),
            "add".to_owned(),
            Some(20),
            None,
            None,
            Some(vec![repo_b]),
        )
        .await;
        assert!(output.is_ok());
        let output = output.ok().unwrap_or_else(|| unreachable!());

        // Both repos have a "src" directory; the grouping must keep them apart.
        let src_groups: Vec<&DirectoryGroup> = output
            .groups
            .iter()
            .filter(|g| g.directory == "src")
            .collect();
        assert!(
            src_groups.len() >= 2,
            "expected two distinct 'src' groups (one per repo), got {}",
            src_groups.len(),
        );
        let repos: HashSet<&str> = src_groups.iter().map(|g| g.repo.as_str()).collect();
        assert!(
            repos.len() >= 2,
            "src groups must come from distinct repos, got: {:?}",
            repos,
        );
    }

    #[tokio::test]
    async fn search_does_not_write_into_extra_repo() {
        let setup = dual_repo_harness().await;
        assert!(setup.is_ok());
        let (fixture_a, fixture_b, _repo_a, repo_b) = setup.ok().unwrap_or_else(|| unreachable!());

        // Snapshot the extra repo's .claudix state-dir tree before the search.
        let state_dir = fixture_b.root().join(".claudix");
        let before = snapshot_paths(&state_dir);

        let output = run_search(
            fixture_a.root(),
            "add".to_owned(),
            Some(10),
            None,
            None,
            Some(vec![repo_b]),
        )
        .await;
        assert!(output.is_ok());

        let after = snapshot_paths(&state_dir);
        assert_eq!(
            before, after,
            "cross-repo search must not write into the extra repo's .claudix dir",
        );
    }

    /// Snapshot every file path and its byte length under `dir` (recursive).
    /// Used to assert read-only behavior: nothing in the snapshot changes.
    fn snapshot_paths(dir: &Path) -> std::collections::BTreeSet<(PathBuf, u64)> {
        fn walk(dir: &Path, into: &mut std::collections::BTreeSet<(PathBuf, u64)>) {
            let Ok(entries) = std::fs::read_dir(dir) else {
                return;
            };
            for entry in entries.flatten() {
                let path = entry.path();
                let Ok(metadata) = entry.metadata() else {
                    continue;
                };
                if metadata.is_dir() {
                    walk(&path, into);
                } else {
                    into.insert((path, metadata.len()));
                }
            }
        }
        let mut set = std::collections::BTreeSet::new();
        walk(dir, &mut set);
        set
    }

    #[test]
    fn effective_cross_repos_orders_config_then_call_args() {
        let cfg = vec!["/cfg/a".to_owned(), "/cfg/b".to_owned()];
        let call = Some(vec!["/cfg/b".to_owned(), "/call/c".to_owned()]);
        let merged = effective_cross_repos(&cfg, call);
        // Config first, call second, exact-string duplicates dropped.
        assert_eq!(merged, vec!["/cfg/a", "/cfg/b", "/call/c"]);
    }

    #[test]
    fn effective_cross_repos_empty_when_both_empty() {
        let merged = effective_cross_repos(&[], None);
        assert!(merged.is_empty());
    }
}