liboxen 0.65.2

Oxen is a fast data version control system, built with machine learning training data in mind. Designed to handle terabytes of data with ease, using a workflow similar to git. Version both structured and unstructured data of any modality: text, images, video, audio, CSV, Parquet, JSONL, model checkpoints, and more. liboxen is the embeddable core library behind the oxen CLI and server, which power fine tuning and inference pipelines for multimodal LLMs, image models, and video models on Oxen.ai.
//! Prune orphaned nodes and version files from the repository
//!
//! This module provides functionality to clean up nodes and version files that are not
//! referenced by any commit in the repository history.

use std::collections::HashSet;

use crate::api;
use crate::error::OxenError;
use crate::model::merkle_tree::node::{EMerkleTreeNode, MerkleTreeNode};
use crate::model::{LocalRepository, MerkleHash, RemoteRepository};
use crate::repositories;

/// Statistics about the prune operation
#[derive(Debug, Clone)]
pub struct PruneStats {
    pub nodes_scanned: usize,
    pub nodes_kept: usize,
    pub nodes_removed: usize,
    pub versions_scanned: usize,
    pub versions_kept: usize,
    pub versions_removed: usize,
    pub bytes_freed: u64,
}

impl PruneStats {
    pub fn new() -> Self {
        Self {
            nodes_scanned: 0,
            nodes_kept: 0,
            nodes_removed: 0,
            versions_scanned: 0,
            versions_kept: 0,
            versions_removed: 0,
            bytes_freed: 0,
        }
    }
}

impl Default for PruneStats {
    fn default() -> Self {
        Self::new()
    }
}

/// Prune orphaned nodes and version files from the repository
///
/// This function:
/// 1. Lists all commits in the repository
/// 2. Traverses all nodes referenced by those commits
/// 3. Collects all version file hashes referenced by file nodes
/// 4. Removes nodes and version files that are not referenced
///
/// # Arguments
/// * `repo` - The local repository to prune
/// * `dry_run` - If true, only report what would be removed without actually removing it
///
/// # Returns
/// Statistics about the prune operation
pub async fn prune(repo: &LocalRepository, dry_run: bool) -> Result<PruneStats, OxenError> {
    let start = std::time::Instant::now();
    let mut stats = PruneStats::new();

    log::info!("Starting prune operation (dry_run: {dry_run})");

    // Step 1: Collect all referenced nodes and version hashes from all commits
    log::info!("Collecting referenced nodes and version files from commit history...");
    let (referenced_nodes, referenced_versions) = collect_referenced_hashes(repo)?;
    log::info!(
        "Found {} referenced nodes and {} referenced version files",
        referenced_nodes.len(),
        referenced_versions.len()
    );

    // Step 2: Prune orphaned nodes
    log::info!("Pruning orphaned nodes...");
    prune_nodes(repo, &referenced_nodes, &mut stats, dry_run)?;

    // Step 3: Prune orphaned version files
    log::info!("Pruning orphaned version files...");
    prune_versions(repo, &referenced_versions, &mut stats, dry_run).await?;

    let duration = start.elapsed();
    log::info!("Prune operation complete in {duration:.2?}");
    log::info!(
        "Nodes: scanned={}, kept={}, removed={}",
        stats.nodes_scanned,
        stats.nodes_kept,
        stats.nodes_removed
    );
    log::info!(
        "Versions: scanned={}, kept={}, removed={}",
        stats.versions_scanned,
        stats.versions_kept,
        stats.versions_removed
    );
    log::info!("Bytes freed: {}", bytesize::ByteSize::b(stats.bytes_freed));

    Ok(stats)
}

/// Prune orphaned nodes and version files from a remote repository
///
/// This function triggers a prune operation on the remote server.
///
/// # Arguments
/// * `remote_repo` - The remote repository to prune
/// * `dry_run` - If true, only report what would be removed without actually removing it
///
/// # Returns
/// Statistics about the prune operation
pub async fn prune_remote(
    remote_repo: &RemoteRepository,
    dry_run: bool,
) -> Result<PruneStats, OxenError> {
    log::info!(
        "Starting remote prune operation on {} (dry_run: {})",
        remote_repo.url(),
        dry_run
    );

    let stats_response = api::client::prune::prune(remote_repo, dry_run).await?;

    // Convert the API response stats to our local PruneStats type
    let stats = PruneStats {
        nodes_scanned: stats_response.nodes_scanned,
        nodes_kept: stats_response.nodes_kept,
        nodes_removed: stats_response.nodes_removed,
        versions_scanned: stats_response.versions_scanned,
        versions_kept: stats_response.versions_kept,
        versions_removed: stats_response.versions_removed,
        bytes_freed: stats_response.bytes_freed,
    };

    log::info!("Remote prune operation complete");
    log::info!(
        "Nodes: scanned={}, kept={}, removed={}",
        stats.nodes_scanned,
        stats.nodes_kept,
        stats.nodes_removed
    );
    log::info!(
        "Versions: scanned={}, kept={}, removed={}",
        stats.versions_scanned,
        stats.versions_kept,
        stats.versions_removed
    );
    log::info!("Bytes freed: {}", bytesize::ByteSize::b(stats.bytes_freed));

    Ok(stats)
}

/// Collect all referenced node hashes and version file hashes from all commits
fn collect_referenced_hashes(
    repo: &LocalRepository,
) -> Result<(HashSet<MerkleHash>, HashSet<String>), OxenError> {
    let mut referenced_nodes = HashSet::new();
    let mut referenced_versions = HashSet::new();

    // Get all commits in the repository
    let all_commits = repositories::commits::list_all(repo)?;
    log::debug!("Processing {} commits", all_commits.len());

    for commit in all_commits {
        // Get the commit node and traverse it
        if let Some(commit_node) = repositories::tree::get_root_with_children(repo, &commit)? {
            collect_node_hashes(
                &commit_node,
                &mut referenced_nodes,
                &mut referenced_versions,
            )?;
        }
    }

    Ok((referenced_nodes, referenced_versions))
}

/// Recursively collect all node hashes and version file hashes from a node tree
fn collect_node_hashes(
    node: &MerkleTreeNode,
    referenced_nodes: &mut HashSet<MerkleHash>,
    referenced_versions: &mut HashSet<String>,
) -> Result<(), OxenError> {
    // Add this node's hash
    referenced_nodes.insert(node.hash);

    // If this is a file node, collect its version hash
    if let EMerkleTreeNode::File(file_node) = &node.node {
        let version_hash = file_node.hash().to_string();
        referenced_versions.insert(version_hash);

        // Also collect chunk hashes if present
        for chunk_hash in file_node.chunk_hashes() {
            referenced_versions.insert(format!("{chunk_hash:032x}"));
        }
    }

    // Recursively process children
    for child in &node.children {
        collect_node_hashes(child, referenced_nodes, referenced_versions)?;
    }

    Ok(())
}

/// Prune orphaned nodes from the repository
fn prune_nodes(
    repo: &LocalRepository,
    referenced_nodes: &HashSet<MerkleHash>,
    stats: &mut PruneStats,
    dry_run: bool,
) -> Result<(), OxenError> {
    // Enumerate and delete through the store so pruning is backend-agnostic, mirroring how
    // `prune_versions` drives the version store below.
    let store = repo.merkle_node_store();
    let all_nodes = store.list_hashes()?;
    stats.nodes_scanned += all_nodes.len();

    for hash in all_nodes {
        if referenced_nodes.contains(&hash) {
            stats.nodes_kept += 1;
        } else {
            // This node is orphaned
            stats.nodes_removed += 1;
            if dry_run {
                log::debug!("Would remove orphaned node: {hash}");
            } else {
                log::debug!("Removing orphaned node: {hash}");
                store.delete(&hash)?;
            }
        }
    }

    Ok(())
}

/// Prune orphaned version files from the repository
async fn prune_versions(
    repo: &LocalRepository,
    referenced_versions: &HashSet<String>,
    stats: &mut PruneStats,
    dry_run: bool,
) -> Result<(), OxenError> {
    let version_store = repo.version_store();

    let all_versions = version_store.list_versions().await?;
    stats.versions_scanned = all_versions.len();

    for version_hash in all_versions {
        if referenced_versions.contains(&version_hash) {
            stats.versions_kept += 1;
        } else {
            // This version is orphaned
            stats.versions_removed += 1;

            // Get the size before deleting
            if let Ok(file_size) = version_store.get_version_size(&version_hash).await {
                stats.bytes_freed += file_size;
            }

            if dry_run {
                log::debug!("Would remove orphaned version: {version_hash}");
            } else {
                log::debug!("Removing orphaned version: {version_hash}");
                version_store.delete_version(&version_hash).await?;
            }
        }
    }

    Ok(())
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::constants::DEFAULT_BRANCH_NAME;
    use crate::test;
    use crate::util::fs::create_dir_all;

    #[tokio::test]
    async fn test_prune_removes_only_unreachable_commits() -> Result<(), OxenError> {
        test::run_empty_local_repo_test_async(|repo| async move {
            let stats = prune(&repo, false).await?;
            assert_eq!(stats.nodes_scanned, 0);
            assert_eq!(stats.nodes_removed, 0);
            assert_eq!(stats.versions_scanned, 0);
            assert_eq!(stats.versions_removed, 0);
            let stats = prune(&repo, true).await?;
            assert_eq!(
                stats.nodes_scanned, 0,
                "a dry run of an empty repo scans nothing"
            );
            assert_eq!(stats.versions_scanned, 0);

            let train_dir = repo.path.join("train");
            create_dir_all(&train_dir)?;
            test::add_txt_file_to_dir(&train_dir, "file1.txt")?;
            test::add_txt_file_to_dir(&train_dir, "file2.txt")?;
            let root_file = repo.path.join("file1.txt");
            tokio::fs::write(&root_file, "file1").await?;
            repositories::add(&repo, &train_dir).await?;
            repositories::add(&repo, &root_file).await?;
            let initial_commit = repositories::commit(&repo, "Add files")?;
            let stats = prune(&repo, false).await?;
            assert_eq!(stats.nodes_removed, 0, "committed data is referenced");
            assert_eq!(stats.versions_removed, 0);

            repositories::branches::create(&repo, "test-branch", &initial_commit.id)?;
            let stats = prune(&repo, false).await?;
            assert_eq!(stats.nodes_removed, 0, "a second branch adds no garbage");
            assert_eq!(stats.versions_removed, 0);

            // A feature branch merged into main and then deleted leaves nothing to remove.
            let branch_name = "feature";
            repositories::branches::create(&repo, branch_name, &initial_commit.id)?;
            repositories::checkout(&repo, branch_name).await?;
            let file3 = repo.path.join("file3.txt");
            tokio::fs::write(&file3, "file3").await?;
            repositories::add(&repo, &file3).await?;
            repositories::commit(&repo, "Add file3")?;
            repositories::checkout(&repo, DEFAULT_BRANCH_NAME).await?;
            repositories::merge::merge(&repo, branch_name).await?;
            repositories::branches::delete(&repo, branch_name)?;

            let commits_before = repositories::commits::list_all(&repo)?;
            let referenced_before = collect_referenced_hashes(&repo)?;
            let stats = prune(&repo, false).await?;
            assert_eq!(
                stats.nodes_removed, 0,
                "merged data is referenced from main"
            );
            assert_eq!(stats.versions_removed, 0);
            assert_eq!(collect_referenced_hashes(&repo)?, referenced_before);
            assert_eq!(repositories::commits::list_all(&repo)?, commits_before);

            // Commits reachable only from deleted branches are removed.
            let head = repositories::commits::head_commit(&repo)?;
            repositories::branches::create(&repo, "branch-1", &head.id)?;
            repositories::checkout(&repo, "branch-1").await?;
            let file = test::add_txt_file_to_dir(&train_dir, "file4.txt")?;
            repositories::add(&repo, &file).await?;
            let branch_1_commit = repositories::commit(&repo, "Commit on branch 1")?;
            repositories::branches::create(&repo, "branch-2", &branch_1_commit.id)?;
            repositories::checkout(&repo, "branch-2").await?;
            let file = test::add_txt_file_to_dir(&train_dir, "file5.txt")?;
            repositories::add(&repo, &file).await?;
            let branch_2_commit = repositories::commit(&repo, "Commit on branch 2")?;
            repositories::checkout(&repo, DEFAULT_BRANCH_NAME).await?;
            repositories::branches::force_delete(&repo, "branch-1")?;
            repositories::branches::force_delete(&repo, "branch-2")?;

            let stats = prune(&repo, true).await?;
            assert_eq!(
                stats.versions_removed, 2,
                "a dry run reports the two branch-only files"
            );
            assert!(stats.nodes_removed > 0);
            for commit in [&branch_1_commit, &branch_2_commit] {
                assert!(
                    repositories::commits::get_by_id(&repo, &commit.id)?.is_some(),
                    "a dry run removes nothing"
                );
            }

            let stats = prune(&repo, false).await?;
            assert_eq!(
                stats.versions_removed, 2,
                "the two branch-only files' versions are removed"
            );
            assert!(stats.nodes_removed > 0);
            for commit in [&branch_1_commit, &branch_2_commit] {
                assert!(
                    repositories::commits::get_by_id(&repo, &commit.id)?.is_none(),
                    "a commit only a deleted branch reached is removed"
                );
            }
            assert_eq!(repositories::commits::list_all(&repo)?, commits_before);

            Ok(())
        })
        .await
    }
}