domani/src/origin/store/git.rs

use crate::error::unexpected::{self, Intoable, Mappable};
use crate::origin::{self, store};

use std::path::{Path, PathBuf};
use std::{collections, fs, io, sync};

#[derive(Clone)]
struct Origin {
    descr: origin::Descr,
    repo: sync::Arc<gix::ThreadSafeRepository>,
    tree_object_id: gix::ObjectId,
}

impl origin::Origin for Origin {
    fn descr(&self) -> &origin::Descr {
        &self.descr
    }

    fn read_file_into(
        &self,
        path: &str,
        into: &mut dyn std::io::Write,
    ) -> Result<(), origin::ReadFileIntoError> {
        let mut clean_path = Path::new(path);
        clean_path = clean_path.strip_prefix("/").unwrap_or(clean_path);

        let repo = self.repo.to_thread_local();

        let file_object = repo
            .find_object(self.tree_object_id)
            .map_unexpected_while(|| format!("finding tree object {}", self.tree_object_id))?
            .peel_to_tree()
            .map_unexpected_while(|| format!("peeling tree object {}", self.tree_object_id))?
            .lookup_entry_by_path(clean_path)
            .map_unexpected_while(|| {
                format!(
                    "looking up {} in tree object {}",
                    clean_path.display(),
                    self.tree_object_id
                )
            })?
            .ok_or(origin::ReadFileIntoError::FileNotFound)?
            .object()
            .or_unexpected()?;

        into.write_all(file_object.data.as_ref())
            .or_unexpected_while("copying out file")?;

        Ok(())
    }
}

#[derive(thiserror::Error, Debug)]
enum GetOriginError {
    #[error("invalid branch name")]
    InvalidBranchName,

    #[error(transparent)]
    Unexpected(#[from] unexpected::Error),
}

/// git::Store implements the Store trait for any Descr::Git based Origins. If any non-git
/// Descrs are used then this implementation will panic.
struct Store {
    dir_path: PathBuf,

    // to prevent against syncing the same origin more than once at a time, but still allowing
    // more than one origin to be syncing at a time
    sync_guard: sync::Mutex<collections::HashMap<origin::Descr, ()>>,

    origins: sync::RwLock<collections::HashMap<origin::Descr, sync::Arc<Origin>>>,
}

pub fn new(dir_path: PathBuf) -> io::Result<sync::Arc<dyn super::Store>> {
    fs::create_dir_all(&dir_path)?;
    Ok(sync::Arc::new(Store {
        dir_path,
        sync_guard: sync::Mutex::new(collections::HashMap::new()),
        origins: sync::RwLock::new(collections::HashMap::new()),
    }))
}

impl Store {
    fn repo_path(&self, descr: &origin::Descr) -> PathBuf {
        self.dir_path.join(descr.id())
    }

    fn descr_file_path(&self, descr_id: &str) -> PathBuf {
        self.dir_path.join(descr_id).join("descr.json")
    }

    fn branch_ref(&self, branch_name: &str) -> String {
        format!("origin/{branch_name}")
    }

    fn get_origin(
        &self,
        repo: gix::Repository,
        descr: origin::Descr,
    ) -> Result<Origin, GetOriginError> {
        let origin::Descr::Git {
            ref branch_name, ..
        } = descr;

        let branch_ref = self.branch_ref(branch_name);

        let commit_object_id = repo
            .try_find_reference(&branch_ref)
            .map_unexpected_while(|| format!("finding branch ref {branch_ref}"))?
            .ok_or(GetOriginError::InvalidBranchName)?
            .peel_to_id_in_place()
            .or_unexpected_while("peeling id in place")?
            .detach();

        let tree_object_id = repo
            .find_object(commit_object_id)
            .map_unexpected_while(|| format!("finding commit object {commit_object_id}"))?
            .try_to_commit_ref()
            .map_unexpected_while(|| format!("parsing {commit_object_id} as commit"))?
            .tree();

        Ok(Origin {
            descr,
            repo: sync::Arc::new(repo.into()),
            tree_object_id,
        })
    }

    fn sync_inner(
        &self,
        descr: &origin::Descr,
        _limits: store::Limits,
    ) -> Result<gix::Repository, store::SyncError> {
        use gix::clone::Error as gixCloneErr;
        use gix::progress::Discard;

        let should_interrupt = &core::sync::atomic::AtomicBool::new(false);
        let repo_path = &self.repo_path(descr);

        // if the path doesn't exist then use the gix clone feature to clone it into the
        // directory.
        if fs::read_dir(repo_path).is_err() {
            fs::create_dir_all(repo_path)
                .map_unexpected_while(|| format!("creating {}", repo_path.display()))?;

            let origin::Descr::Git {
                ref url,
                ref branch_name,
            } = descr;

            let (repo, _) = gix::prepare_clone_bare(url.clone(), repo_path)
                .map_err(|e| match e {
                    gixCloneErr::Init(gix::init::Error::InvalidBranchName { .. }) => {
                        store::SyncError::InvalidBranchName
                    }
                    gixCloneErr::UrlParse(_) | gixCloneErr::CanonicalizeUrl { .. } => {
                        store::SyncError::InvalidURL
                    }
                    _ => e
                        .into_unexpected_while(format!(
                            "cloning {} into {}",
                            url,
                            repo_path.display()
                        ))
                        .into(),
                })?
                .fetch_only(Discard, should_interrupt)
                .map_err(|_| store::SyncError::InvalidURL)?;

            // Check to make sure the branch name exists
            // TODO if this fails we should delete repo_path
            let branch_ref = self.branch_ref(branch_name);
            repo.try_find_reference(&branch_ref)
                .map_unexpected_while(|| format!("finding branch ref {branch_ref}"))?
                .ok_or(store::SyncError::InvalidBranchName)?;

            // Add the descr to the repo directory, so we can know the actual descr later
            // TODO if this fails we should delete repo_path
            let file_path = self.descr_file_path(descr.id().as_ref());
            let descr_file = fs::File::create(&file_path)
                .map_unexpected_while(|| format!("creating {}", file_path.display()))?;

            serde_json::to_writer(descr_file, &descr)
                .map_unexpected_while(|| format!("writing descr to {}", file_path.display()))?;

            return Ok(repo);
        }

        let direction = gix::remote::Direction::Fetch;

        let repo = gix::open(repo_path)
            .map_unexpected_while(|| format!("opening repo at {}", repo_path.display()))?;

        let remote = repo
            .find_default_remote(direction)
            .ok_or_else(|| unexpected::Error::from("no default configured"))?
            .or_unexpected_while("finding default remote for fetching")?;

        remote
            .connect(direction)
            .or_unexpected_while("connecting to remote")?
            .prepare_fetch(Discard, Default::default())
            .or_unexpected_while("preparing fetch")?
            .receive(Discard, should_interrupt)
            .or_unexpected_while("fetching from remote")?;

        Ok(repo)
    }
}

impl super::Store for Store {
    fn sync(&self, descr: origin::Descr, limits: store::Limits) -> Result<(), store::SyncError> {
        // attempt to lock this descr for syncing, doing so within a new scope so the mutex
        // isn't actually being held for the whole method duration.
        let is_already_syncing = {
            self.sync_guard
                .lock()
                .unwrap()
                .insert(descr.clone(), ())
                .is_some()
        };

        if is_already_syncing {
            return Err(store::SyncError::AlreadyInProgress);
        }

        let res = self.sync_inner(&descr, limits);

        self.sync_guard.lock().unwrap().remove(&descr);

        let repo = match res {
            Ok(repo) => repo,
            Err(e) => return Err(e),
        };

        // repo is synced at this point (though the sync lock is still held), just gotta create
        // the origin and store it.
        //
        // TODO this is a bit of a memory leak, but by the time we get
        // to that point this should all be backed by something which isn't local storage
        // anyway.

        // calling this while the sync lock is held isn't ideal, but it's convenient and
        // shouldn't be too terrible generally
        let origin = self.get_origin(repo, descr.clone()).map_err(|e| match e {
            GetOriginError::InvalidBranchName => store::SyncError::InvalidBranchName,
            GetOriginError::Unexpected(e) => store::SyncError::Unexpected(e),
        })?;

        let mut origins = self.origins.write().unwrap();
        (*origins).insert(descr, sync::Arc::new(origin));

        Ok(())
    }

    fn get(&self, descr: origin::Descr) -> Result<sync::Arc<dyn origin::Origin>, store::GetError> {
        {
            let origins = self.origins.read().unwrap();
            if let Some(origin) = origins.get(&descr) {
                return Ok(origin.clone());
            }
        }

        let repo_path = self.repo_path(&descr);

        fs::read_dir(&repo_path).map_err(|e| match e.kind() {
            io::ErrorKind::NotFound => store::GetError::NotFound,
            _ => e
                .into_unexpected_while(format!("checking if {} exists", repo_path.display()))
                .into(),
        })?;

        let repo = gix::open(&repo_path)
            .map_unexpected_while(|| format!("opening {} as git repo", repo_path.display()))?;

        let origin = self.get_origin(repo, descr.clone()).map_err(|e| match e {
            // it's not expected that the branch name is invalid at this point, it must have
            // existed for sync to have been successful.
            GetOriginError::InvalidBranchName => e.into_unexpected().into(),
            GetOriginError::Unexpected(e) => store::GetError::Unexpected(e),
        })?;

        let origin = sync::Arc::new(origin.clone());

        let mut origins = self.origins.write().unwrap();

        (*origins).insert(descr, origin.clone());

        Ok(origin)
    }

    fn all_descrs(&self) -> Result<Vec<origin::Descr>, store::AllDescrsError> {
        fs::read_dir(&self.dir_path).or_unexpected()?.map(
                |dir_entry_res: io::Result<fs::DirEntry>| -> Result<origin::Descr, store::AllDescrsError> {
                    let descr_id: String = dir_entry_res
                        .or_unexpected()?
                        .file_name()
                        .to_str()
                        .ok_or_else(|| {
                            unexpected::Error::from("couldn't convert os string to &str")
                        })?
                        .into();

                    let descr_file_path = self.descr_file_path(descr_id.as_ref());

                    // TODO it's possible that opening the file will fail if syncing is
                    // still ongoing, as writing the descr file is the last step after
                    // initial sync has succeeded.
                    let descr_file = fs::File::open(descr_file_path.as_path())
                        .map_unexpected_while(|| {
                            format!("opening descr file {}", descr_file_path.display())
                        })?;

                    let descr = serde_json::from_reader(descr_file).map_unexpected_while(|| {
                        format!("reading descr file {}", descr_file_path.display())
                    })?;

                    Ok(descr)
                },
            ).try_collect()
    }
}

#[cfg(test)]
mod tests {
    use crate::origin::store;
    use crate::origin::store::Store;
    use crate::origin::{self, Origin};
    use tempdir::TempDir;

    #[test]
    fn basic() {
        let tmp_dir = TempDir::new("origin_store_git").unwrap();

        let curr_dir = format!("file://{}", std::env::current_dir().unwrap().display());

        let descr = origin::Descr::Git {
            url: curr_dir.clone(),
            branch_name: String::from("main"),
        };

        let other_descr = origin::Descr::Git {
            url: curr_dir.clone(),
            branch_name: String::from("some_other_branch"),
        };

        let limits = store::Limits {};

        let store = super::new(tmp_dir.path().to_path_buf()).expect("store created");

        store
            .sync(descr.clone(), limits)
            .expect("sync should succeed");
        store
            .sync(descr.clone(), limits)
            .expect("second sync should succeed");

        assert!(matches!(
            store.get(other_descr),
            Err::<_, store::GetError>(store::GetError::NotFound),
        ));

        let origin = store.get(descr.clone()).expect("origin retrieved");

        assert_eq!(&descr, origin.descr());

        let assert_write = |path: &str| {
            let mut into: Vec<u8> = vec![];
            origin
                .read_file_into(path, &mut into)
                .expect("write should succeed");
            assert!(into.len() > 0);
        };

        assert_write("src/lib.rs");
        assert_write("/src/lib.rs");

        // File doesn't exist
        let mut into: Vec<u8> = vec![];
        assert!(matches!(
            origin.read_file_into("DNE", &mut into),
            Err::<(), origin::ReadFileIntoError>(origin::ReadFileIntoError::FileNotFound),
        ));
        assert_eq!(into.len(), 0);

        let descrs = store
            .all_descrs()
            .expect("all_descrs called")
            .into_iter()
            .collect::<Vec<Result<origin::Descr, store::AllDescrsError>>>();

        assert_eq!(1, descrs.len());
        assert_eq!(&descr, descrs[0].as_ref().unwrap());
    }
}