diff options
Diffstat (limited to 'crates/daemon/src/aclient')
| -rw-r--r-- | crates/daemon/src/aclient/api_client.rs | 201 | ||||
| -rw-r--r-- | crates/daemon/src/aclient/database/mod.rs | 214 | ||||
| -rw-r--r-- | crates/daemon/src/aclient/encryption.rs | 142 | ||||
| -rw-r--r-- | crates/daemon/src/aclient/history/mod.rs | 374 | ||||
| -rw-r--r-- | crates/daemon/src/aclient/history/store.rs | 296 | ||||
| -rw-r--r-- | crates/daemon/src/aclient/meta.rs | 182 | ||||
| -rw-r--r-- | crates/daemon/src/aclient/mod.rs | 9 | ||||
| -rw-r--r-- | crates/daemon/src/aclient/record/encryption.rs | 379 | ||||
| -rw-r--r-- | crates/daemon/src/aclient/record/mod.rs | 3 | ||||
| -rw-r--r-- | crates/daemon/src/aclient/record/sqlite_store.rs | 412 | ||||
| -rw-r--r-- | crates/daemon/src/aclient/record/sync.rs | 453 | ||||
| -rw-r--r-- | crates/daemon/src/aclient/settings/meta.rs | 17 | ||||
| -rw-r--r-- | crates/daemon/src/aclient/settings/mod.rs | 590 | ||||
| -rw-r--r-- | crates/daemon/src/aclient/utils.rs | 77 |
14 files changed, 3349 insertions, 0 deletions
diff --git a/crates/daemon/src/aclient/api_client.rs b/crates/daemon/src/aclient/api_client.rs new file mode 100644 index 00000000..1eba51bd --- /dev/null +++ b/crates/daemon/src/aclient/api_client.rs @@ -0,0 +1,201 @@ +use std::env; +use std::time::Duration; + +use eyre::{Result, bail, eyre}; +use reqwest::{Response, StatusCode, Url, header::HeaderMap}; +use tracing::debug; +use uuid::Uuid; + +use turtle_common::{api::ErrorResponse, record::RecordStatus}; +use turtle_common::{ + api::{ATUIN_CARGO_VERSION, ATUIN_HEADER_VERSION, ATUIN_VERSION}, + record::{EncryptedData, HostId, Record, RecordIdx}, +}; + +use semver::Version; + +static APP_USER_AGENT: &str = concat!("atuin/", env!("CARGO_PKG_VERSION"),); + +pub(crate) struct Client<'a> { + sync_addr: &'a str, + user_id: Uuid, + inner: reqwest::Client, +} + +fn make_url(address: &str, path: &str, user_id: Uuid) -> Result<String> { + let address = address.strip_suffix('/').unwrap_or(address); + + // `join()` expects a trailing `/` in order to join paths + // e.g. it treats `http://host:port/subdir` as a file called `subdir` + let address = &format!("{address}/api/v0/{user_id}/"); + + // passing a path with a leading `/` will cause `join()` to replace the entire URL path + let path = path.strip_prefix("/").unwrap_or(path); + + let url = Url::parse(address) + .map(|url| url.join(path))? + .map_err(|_| eyre!("invalid address"))?; + + Ok(url.to_string()) +} + +fn ensure_version(response: &Response) -> Result<bool> { + let version = response.headers().get(ATUIN_HEADER_VERSION); + + let version = if let Some(version) = version { + match version.to_str() { + Ok(v) => Version::parse(v), + Err(e) => bail!("failed to parse server version: {:?}", e), + } + } else { + bail!("Server not reporting its version: it is either too old or unhealthy"); + }?; + + // If the client is newer than the server + if version.major < ATUIN_VERSION.major { + println!( + "Atuin version mismatch! In order to successfully sync, the server needs to run a newer version of Atuin" + ); + println!("Client: {ATUIN_CARGO_VERSION}"); + println!("Server: {version}"); + + return Ok(false); + } + + Ok(true) +} + +async fn handle_resp_error(resp: Response) -> Result<Response> { + let status = resp.status(); + let url = resp.url().to_string(); + + if status == StatusCode::SERVICE_UNAVAILABLE { + bail!( + "Service unavailable: check https://status.atuin.sh (or get in touch with your host)" + ); + } + + if status == StatusCode::TOO_MANY_REQUESTS { + bail!("Rate limited; please wait before doing that again"); + } + + if !status.is_success() { + if let Ok(error) = resp.json::<ErrorResponse<'_>>().await { + let reason = error.reason; + + if status.is_client_error() { + bail!("Invalid request to the service at {url}, {status} - {reason}.") + } + + bail!( + "There was an error with the atuin sync service at {url}, server error {status}: {reason}.\nIf the problem persists, contact the host" + ) + } + + bail!( + "There was an error with the atuin sync service at {url}, Status {status:?}.\nIf the problem persists, contact the host" + ) + } + + Ok(resp) +} + +use std::sync::Once; + +static INIT: Once = Once::new(); + +/// Ensure the rustls crypto provider (ring) is installed. +/// +/// Must be called before creating any reqwest clients. Safe to call +/// multiple times — only the first call installs the provider. +fn ensure_crypto_provider() { + INIT.call_once(|| { + rustls::crypto::ring::default_provider() + .install_default() + .expect("Failed to install rustls crypto provider"); + }); +} + +impl<'a> Client<'a> { + pub(crate) fn new( + sync_addr: &'a str, + connect_timeout: u64, + timeout: u64, + user_id: Uuid, + ) -> Result<Self> { + ensure_crypto_provider(); + let mut headers = HeaderMap::new(); + + // used for semver server check + headers.insert(ATUIN_HEADER_VERSION, ATUIN_CARGO_VERSION.parse()?); + + Ok(Client { + user_id, + sync_addr, + inner: reqwest::Client::builder() + .user_agent(APP_USER_AGENT) + .default_headers(headers) + .connect_timeout(Duration::new(connect_timeout, 0)) + .timeout(Duration::new(timeout, 0)) + .build()?, + }) + } + + pub(crate) async fn post_records(&self, records: &[Record<EncryptedData>]) -> Result<()> { + let url = make_url(self.sync_addr, "/record", self.user_id)?; + let url = Url::parse(url.as_str())?; + + debug!("uploading {} records to {url}", records.len()); + + let resp = self.inner.post(url).json(records).send().await?; + handle_resp_error(resp).await?; + + Ok(()) + } + + pub(crate) async fn next_records( + &self, + host: HostId, + tag: String, + start: RecordIdx, + count: u64, + ) -> Result<Vec<Record<EncryptedData>>> { + debug!("fetching record/s from host {}/{}/{}", host.0, tag, start); + + let url = make_url( + self.sync_addr, + &format!( + "/record/next?host={}&tag={}&count={}&start={}", + host.0, tag, count, start + ), + self.user_id, + )?; + + let url = Url::parse(url.as_str())?; + + let resp = self.inner.get(url).send().await?; + let resp = handle_resp_error(resp).await?; + + let records = resp.json::<Vec<Record<EncryptedData>>>().await?; + + Ok(records) + } + + pub(crate) async fn record_status(&self) -> Result<RecordStatus> { + let url = make_url(self.sync_addr, "/record", self.user_id)?; + let url = Url::parse(url.as_str())?; + + let resp = self.inner.get(url).send().await?; + let resp = handle_resp_error(resp).await?; + + if !ensure_version(&resp)? { + bail!("could not sync records due to version mismatch"); + } + + let index = resp.json().await?; + + debug!("got remote index {index:?}"); + + Ok(index) + } +} diff --git a/crates/daemon/src/aclient/database/mod.rs b/crates/daemon/src/aclient/database/mod.rs new file mode 100644 index 00000000..7cee866e --- /dev/null +++ b/crates/daemon/src/aclient/database/mod.rs @@ -0,0 +1,214 @@ +use std::{path::Path, str::FromStr, time::Duration}; + +use fs_err::{self as fs}; +use sql_builder::{SqlBuilder, SqlName}; +use sqlx::{ + AssertSqlSafe, Result, Row, + sqlite::{SqliteConnectOptions, SqliteJournalMode, SqlitePool, SqliteRow, SqliteSynchronous}, +}; +use time::OffsetDateTime; +use tracing::debug; +use turtle_api::history::{History, HistoryId}; +use turtle_common::utils; + +use crate::aclient::utils::setup_db; + +// Intended for use on a developer machine and not a sync server. +// TODO: implement IntoIterator +#[derive(Debug, Clone)] +pub(crate) struct ClientSqlite { + pool: SqlitePool, +} + +impl ClientSqlite { + pub(crate) async fn new(path: impl AsRef<Path>, timeout: f64) -> Result<Self> { + fn mk_opts(path: &str) -> Result<SqliteConnectOptions> { + let opts = SqliteConnectOptions::from_str(path)? + .journal_mode(SqliteJournalMode::Wal) + .optimize_on_close(true, None) + .synchronous(SqliteSynchronous::Normal) + .with_regexp() + .create_if_missing(true); + + Ok(opts) + } + + let path = path.as_ref(); + debug!("opening sqlite database at {path:?}"); + + if utils::broken_symlink(path) { + eprintln!( + "Atuin: Sqlite db path ({}) is a broken symlink. Unable to read or create replacement.", + path.display() + ); + std::process::exit(1); + } + + if !path.exists() + && let Some(dir) = path.parent() + { + fs::create_dir_all(dir)?; + } + + let pool = setup_db!(path, timeout, mk_opts, "./db/client-migrations").await?; + Ok(Self { pool }) + } + + async fn save_raw(tx: &mut sqlx::Transaction<'_, sqlx::Sqlite>, h: &History) -> Result<()> { + sqlx::query( + " + INSERT OR IGNORE + INTO history (id, timestamp, duration, exit, command, cwd, session, hostname, author, intent, deleted_at) + VALUES (?1, ?2, ?3, ?4, ?5, ?6, ?7, ?8, ?9, ?10, ?11) + ", + ) + .bind(h.id.to_string().as_str()) + .bind(h.timestamp.unix_timestamp_nanos() as i64) + .bind(h.duration.as_nanos() as i64) + .bind(h.exit) + .bind(h.command.as_str()) + .bind(h.cwd.as_str()) + .bind(h.session.as_str()) + .bind(h.hostname.as_str()) + .bind(h.author.as_str()) + .bind(h.intent.as_deref()) + .bind(h.deleted_at.map(|t|t.unix_timestamp_nanos() as i64)) + .execute(&mut **tx) + .await?; + + Ok(()) + } + + async fn delete_row_raw( + tx: &mut sqlx::Transaction<'_, sqlx::Sqlite>, + id: HistoryId, + ) -> Result<()> { + sqlx::query("delete from history where id = ?1") + .bind(id.to_string().as_str()) + .execute(&mut **tx) + .await?; + + Ok(()) + } + + #[expect(clippy::needless_pass_by_value)] + fn query_history_inner(row: SqliteRow) -> History { + let deleted_at: Option<i64> = row.get("deleted_at"); + let hostname: String = row.get("hostname"); + let author: Option<String> = row.try_get("author").ok().flatten(); + let author = author + .filter(|author| !author.trim().is_empty()) + .unwrap_or_else(|| History::author_from_hostname(hostname.as_str())); + let intent: Option<String> = row.try_get("intent").ok().flatten(); + let intent = intent.filter(|intent| !intent.trim().is_empty()); + + History::from_db() + .id(row.get("id")) + .timestamp( + OffsetDateTime::from_unix_timestamp_nanos(i128::from( + row.get::<i64, _>("timestamp"), + )) + .unwrap(), + ) + .duration(Duration::from_nanos( + u64::try_from(row.get::<i64, _>("duration")).expect("to be small enough"), + )) + .exit(row.get("exit")) + .command(row.get("command")) + .cwd(row.get("cwd")) + .session(row.get("session")) + .hostname(hostname) + .author(author) + .intent(intent) + .deleted_at( + deleted_at + .and_then(|t| OffsetDateTime::from_unix_timestamp_nanos(i128::from(t)).ok()), + ) + .build() + .into() + } +} + +impl ClientSqlite { + pub(crate) async fn save(&self, h: &History) -> Result<()> { + debug!("saving history to sqlite"); + let mut tx = self.pool.begin().await?; + Self::save_raw(&mut tx, h).await?; + tx.commit().await?; + + Ok(()) + } + + /// make a unique list, that only shows the *newest* version of things + pub(crate) async fn list( + &self, + max: Option<usize>, + unique: bool, + include_deleted: bool, + ) -> Result<Vec<History>> { + debug!("listing history"); + + let mut query = SqlBuilder::select_from(SqlName::new("history").alias("h").baquoted()); + query.field("*").order_desc("timestamp"); + if !include_deleted { + query.and_where_is_null("deleted_at"); + } + + if unique { + query.group_by("command").having("max(timestamp)"); + } + + if let Some(max) = max { + let max: usize = max; + query.limit(max); + } + + let query = query.sql().expect("bug in list query. please report"); + + // SAFETY: + // - The query is constructed via sql_bulider, and as such should be safe. + // - The only value, that is directly added to the query is a `usize`. + let res = sqlx::query(AssertSqlSafe(query)) + .map(Self::query_history_inner) + .fetch_all(&self.pool) + .await?; + + Ok(res) + } + + pub(crate) async fn range( + &self, + from: OffsetDateTime, + to: OffsetDateTime, + ) -> Result<Vec<History>> { + debug!("listing history from {:?} to {:?}", from, to); + + let res = sqlx::query( + " + SELECT * + FROM history + WHERE timestamp >= ?1 AND timestamp <= ?2 + ORDER BY timestamp ASC + ", + ) + .bind(from.unix_timestamp_nanos() as i64) + .bind(to.unix_timestamp_nanos() as i64) + .map(Self::query_history_inner) + .fetch_all(&self.pool) + .await?; + + Ok(res) + } + + pub(crate) async fn delete_rows(&self, ids: &[HistoryId]) -> Result<()> { + let mut tx = self.pool.begin().await?; + + for id in ids { + Self::delete_row_raw(&mut tx, id.clone()).await?; + } + + tx.commit().await?; + + Ok(()) + } +} diff --git a/crates/daemon/src/aclient/encryption.rs b/crates/daemon/src/aclient/encryption.rs new file mode 100644 index 00000000..45e82ab3 --- /dev/null +++ b/crates/daemon/src/aclient/encryption.rs @@ -0,0 +1,142 @@ +// The general idea is that we NEVER send cleartext history to the server +// This way the odds of anything private ending up where it should not are +// very low +// The server authenticates via the usual username and password. This has +// nothing to do with the encryption, and is purely authentication! The client +// generates its own secret key, and encrypts all shell history with libsodium's +// secretbox. The data is then sent to the server, where it is stored. All +// clients must share the secret in order to be able to sync, as it is needed +// to decrypt + +use std::io::prelude::Write; + +use base64::prelude::{BASE64_STANDARD, Engine}; +use crypto_secretbox::Key; +use crypto_secretbox::{KeyInit, XSalsa20Poly1305, aead::OsRng}; +use eyre::{Context, Result, bail, ensure, eyre}; +use fs_err as fs; +use rmp::Marker; + +use crate::aclient::settings::Settings; + +fn generate_encoded_key() -> Result<(Key, String)> { + let key = XSalsa20Poly1305::generate_key(&mut OsRng); + let encoded = encode_key(&key)?; + + Ok((key, encoded)) +} + +fn new_key(settings: &Settings) -> Result<Key> { + if settings.sync.encryption_key()?.is_some() { + bail!("key already exists! cannot overwrite"); + } else if let Some(path) = settings.sync.encryption_key_path.as_ref() { + let (key, encoded) = generate_encoded_key()?; + + let mut file = fs::File::create(path)?; + file.write_all(encoded.as_bytes())?; + + Ok(key) + } else { + bail!("No key-path set, cannot generate key") + } +} + +// Loads the secret key, will create + save if it doesn't exist +pub(crate) fn load_key(settings: &Settings) -> Result<Key> { + if let Some(key) = settings.sync.encryption_key()? { + Ok(key) + } else { + Ok(new_key(settings)?) + } +} + +fn encode_key(key: &Key) -> Result<String> { + let mut buf = vec![]; + rmp::encode::write_array_len(&mut buf, key.len() as u32) + .wrap_err("could not encode key to message pack")?; + for b in key { + rmp::encode::write_uint(&mut buf, u64::from(*b)) + .wrap_err("could not encode key to message pack")?; + } + let buf = BASE64_STANDARD.encode(buf); + + Ok(buf) +} + +pub(crate) fn decode_key(key: &str) -> Result<Key> { + use rmp::decode; + + let buf = BASE64_STANDARD + .decode(key.trim_end()) + .wrap_err("encryption key is not a valid base64 encoding")?; + + // old code wrote the key as a fixed length array of 32 bytes + // new code writes the key with a length prefix + if let Ok(key) = <[u8; 32]>::try_from(&*buf) { + Ok(key.into()) + } else { + let mut bytes = decode::Bytes::new(&buf); + + match Marker::from_u8(buf[0]) { + Marker::Bin8 => { + let len = decode::read_bin_len(&mut bytes).map_err(|err| eyre!("{err:?}"))?; + ensure!(len == 32, "encryption key is not the correct size"); + let key = <[u8; 32]>::try_from(bytes.remaining_slice()) + .context("could not decode encryption key")?; + Ok(key.into()) + } + Marker::Array16 => { + let len = decode::read_array_len(&mut bytes).map_err(|err| eyre!("{err:?}"))?; + ensure!(len == 32, "encryption key is not the correct size"); + + let mut key = Key::default(); + for i in &mut key { + *i = decode::read_int(&mut bytes).map_err(|err| eyre!("{err:?}"))?; + } + Ok(key) + } + _ => bail!("could not decode encryption key"), + } + } +} + +#[cfg(test)] +mod test { + #[test] + fn key_encodings() { + use super::{Key, decode_key, encode_key}; + + // a history of our key encodings. + // v11.0.0 xCAbWypb0msJ2Kq+8j4GVEWUlDX7deKnrTRSIopuqXxc5Q== + // v12.0.0 xCAbWypb0msJ2Kq+8j4GVEWUlDX7deKnrTRSIopuqXxc5Q== + // v13.0.0 xCAbWypb0msJ2Kq+8j4GVEWUlDX7deKnrTRSIopuqXxc5Q== + // v13.0.1 xCAbWypb0msJ2Kq+8j4GVEWUlDX7deKnrTRSIopuqXxc5Q== + // v14.0.0 xCAbWypb0msJ2Kq+8j4GVEWUlDX7deKnrTRSIopuqXxc5Q== + // v14.0.1 xCAbWypb0msJ2Kq+8j4GVEWUlDX7deKnrTRSIopuqXxc5Q== + // c7d89c1 3AAgG1sqW8zSawnM2MyqzL7M8j4GVEXMlMyUNcz7dczizKfMrTRSIsyKbsypfFzM5Q== (https://github.com/ellie/atuin/pull/805) + // b53ca35 3AAgG1sqW8zSawnM2MyqzL7M8j4GVEXMlMyUNcz7dczizKfMrTRSIsyKbsypfFzM5Q== (https://github.com/ellie/atuin/pull/974) + // v15.0.0 3AAgG1sqW8zSawnM2MyqzL7M8j4GVEXMlMyUNcz7dczizKfMrTRSIsyKbsypfFzM5Q== + // b8b57c8 xCAbWypb0msJ2Kq+8j4GVEWUlDX7deKnrTRSIopuqXxc5Q== (https://github.com/ellie/atuin/pull/1057) + // 8c94d79 3AAgG1sqW8zSawnM2MyqzL7M8j4GVEXMlMyUNcz7dczizKfMrTRSIsyKbsypfFzM5Q== (https://github.com/ellie/atuin/pull/1089) + + let key = Key::from([ + 27, 91, 42, 91, 210, 107, 9, 216, 170, 190, 242, 62, 6, 84, 69, 148, 148, 53, 251, 117, + 226, 167, 173, 52, 82, 34, 138, 110, 169, 124, 92, 229, + ]); + + assert_eq!( + encode_key(&key).unwrap(), + "3AAgG1sqW8zSawnM2MyqzL7M8j4GVEXMlMyUNcz7dczizKfMrTRSIsyKbsypfFzM5Q==" + ); + + // key encodings we have to support + let valid_encodings = [ + "xCAbWypb0msJ2Kq+8j4GVEWUlDX7deKnrTRSIopuqXxc5Q==", + "3AAgG1sqW8zSawnM2MyqzL7M8j4GVEXMlMyUNcz7dczizKfMrTRSIsyKbsypfFzM5Q==", + ]; + + for k in valid_encodings { + assert_eq!(decode_key(k).expect(k), key); + } + } +} diff --git a/crates/daemon/src/aclient/history/mod.rs b/crates/daemon/src/aclient/history/mod.rs new file mode 100644 index 00000000..0b5f4982 --- /dev/null +++ b/crates/daemon/src/aclient/history/mod.rs @@ -0,0 +1,374 @@ +use std::time::Duration; + +use rmp::decode::DecodeStringError; +use rmp::decode::ValueReadError; +use rmp::{Marker, decode::Bytes}; +use turtle_api::history::History; + +use turtle_common::record::DecryptedData; + +use eyre::{Result, bail, eyre}; + +use time::OffsetDateTime; + +pub(crate) mod store; + +const HISTORY_VERSION_V0: &str = "v0"; +const HISTORY_VERSION_V1: &str = "v1"; +const HISTORY_RECORD_VERSION_V0: u16 = 0; +const HISTORY_RECORD_VERSION_V1: u16 = 1; +const HISTORY_VERSION: &str = HISTORY_VERSION_V1; +const HISTORY_TAG: &str = "history"; + +trait HistoryExt: Sized { + fn serialize(&self) -> Result<DecryptedData>; + fn read_optional_string(bytes: &[u8]) -> Result<(Option<String>, &[u8])>; + fn deserialize_v0(bytes: &[u8]) -> Result<Self>; + fn deserialize_v1(bytes: &[u8]) -> Result<Self>; + fn deserialize(bytes: &[u8], version: &str) -> Result<Self>; +} + +impl HistoryExt for History { + fn serialize(&self) -> Result<DecryptedData> { + // This is pretty much the same as what we used for the old history, with one difference - + // it uses integers for timestamps rather than a string format. + + use rmp::encode; + + let mut output = vec![]; + + // write the version + encode::write_u16(&mut output, HISTORY_RECORD_VERSION_V1)?; + let include_intent = self.intent.is_some(); + encode::write_array_len(&mut output, 10 + u32::from(include_intent))?; + + encode::write_str(&mut output, &self.id.to_string())?; + encode::write_u64(&mut output, self.timestamp.unix_timestamp_nanos() as u64)?; + encode::write_sint( + &mut output, + i64::try_from(self.duration.as_nanos()).expect("should be small enough"), + )?; + encode::write_sint(&mut output, self.exit)?; + encode::write_str(&mut output, &self.command)?; + encode::write_str(&mut output, &self.cwd)?; + encode::write_str(&mut output, &self.session)?; + encode::write_str(&mut output, &self.hostname)?; + + match self.deleted_at { + Some(d) => encode::write_u64(&mut output, d.unix_timestamp_nanos() as u64)?, + None => encode::write_nil(&mut output)?, + } + + encode::write_str(&mut output, self.author.as_str())?; + if let Some(intent) = &self.intent { + encode::write_str(&mut output, intent.as_str())?; + } + + Ok(DecryptedData(output)) + } + + fn read_optional_string(bytes: &[u8]) -> Result<(Option<String>, &[u8])> { + use rmp::decode; + + fn error_report<E: std::fmt::Debug>(err: E) -> eyre::Report { + eyre!("{err:?}") + } + + match decode::read_str_from_slice(bytes) { + Ok((value, bytes)) => Ok((Some(value.to_owned()), bytes)), + Err(DecodeStringError::TypeMismatch(Marker::Null)) => { + let mut cursor = Bytes::new(bytes); + decode::read_nil(&mut cursor).map_err(error_report)?; + + Ok((None, cursor.remaining_slice())) + } + Err(err) => Err(error_report(err)), + } + } + + fn deserialize_v0(bytes: &[u8]) -> Result<Self> { + use rmp::decode; + + fn error_report<E: std::fmt::Debug>(err: E) -> eyre::Report { + eyre!("{err:?}") + } + + let mut bytes = Bytes::new(bytes); + + let version = decode::read_u16(&mut bytes).map_err(error_report)?; + + if version != HISTORY_RECORD_VERSION_V0 { + bail!("expected decoding v0 record, found v{version}"); + } + + let nfields = decode::read_array_len(&mut bytes).map_err(error_report)?; + + if nfields != 9 { + bail!("cannot decrypt history from a different version of Atuin"); + } + + let bytes = bytes.remaining_slice(); + let (id, bytes) = decode::read_str_from_slice(bytes).map_err(error_report)?; + + let mut bytes = Bytes::new(bytes); + let timestamp = decode::read_u64(&mut bytes).map_err(error_report)?; + let duration = decode::read_int(&mut bytes) + .map(|int: i64| { + Duration::from_nanos(u64::try_from(int).expect("should be small enough")) + }) + .map_err(error_report)?; + let exit = decode::read_int(&mut bytes).map_err(error_report)?; + + let bytes = bytes.remaining_slice(); + let (command, bytes) = decode::read_str_from_slice(bytes).map_err(error_report)?; + let (cwd, bytes) = decode::read_str_from_slice(bytes).map_err(error_report)?; + let (session, bytes) = decode::read_str_from_slice(bytes).map_err(error_report)?; + let (hostname, bytes) = decode::read_str_from_slice(bytes).map_err(error_report)?; + + let mut bytes = Bytes::new(bytes); + + let (deleted_at, bytes) = match decode::read_u64(&mut bytes) { + Ok(unix) => (Some(unix), bytes.remaining_slice()), + // we accept null here + Err(ValueReadError::TypeMismatch(Marker::Null)) => (None, bytes.remaining_slice()), + Err(err) => return Err(error_report(err)), + }; + if !bytes.is_empty() { + bail!("trailing bytes in encoded history. malformed") + } + + Ok(Self { + id: id.to_owned().into(), + timestamp: OffsetDateTime::from_unix_timestamp_nanos(i128::from(timestamp))?, + duration, + exit, + command: command.to_owned(), + cwd: cwd.to_owned(), + session: session.to_owned(), + hostname: hostname.to_owned(), + author: Self::author_from_hostname(hostname), + intent: None, + deleted_at: deleted_at + .map(|t| OffsetDateTime::from_unix_timestamp_nanos(i128::from(t))) + .transpose()?, + }) + } + + fn deserialize_v1(bytes: &[u8]) -> Result<Self> { + use rmp::decode; + + fn error_report<E: std::fmt::Debug>(err: E) -> eyre::Report { + eyre!("{err:?}") + } + + let mut bytes = Bytes::new(bytes); + + let version = decode::read_u16(&mut bytes).map_err(error_report)?; + + if version != HISTORY_RECORD_VERSION_V1 { + bail!("expected decoding v1 record, found v{version}"); + } + + let nfields = decode::read_array_len(&mut bytes).map_err(error_report)?; + + if !(10..=11).contains(&nfields) { + bail!("cannot decrypt history from a different version of Atuin"); + } + + let bytes = bytes.remaining_slice(); + let (id, bytes) = decode::read_str_from_slice(bytes).map_err(error_report)?; + + let mut bytes = Bytes::new(bytes); + let timestamp = decode::read_u64(&mut bytes).map_err(error_report)?; + let duration = decode::read_int(&mut bytes) + .map(|int: i64| { + if int.is_negative() { + // TODO: We should probably handle this case differently <2026-08-23> + Duration::from_nanos(0) + } else { + Duration::from_nanos(u64::try_from(int).expect("to be small enough")) + } + }) + .map_err(error_report)?; + let exit = decode::read_int(&mut bytes).map_err(error_report)?; + + let bytes = bytes.remaining_slice(); + let (command, bytes) = decode::read_str_from_slice(bytes).map_err(error_report)?; + let (cwd, bytes) = decode::read_str_from_slice(bytes).map_err(error_report)?; + let (session, bytes) = decode::read_str_from_slice(bytes).map_err(error_report)?; + let (hostname, bytes) = decode::read_str_from_slice(bytes).map_err(error_report)?; + + let mut bytes = Bytes::new(bytes); + + let (deleted_at, bytes) = match decode::read_u64(&mut bytes) { + Ok(unix) => (Some(unix), bytes.remaining_slice()), + // we accept null here + Err(ValueReadError::TypeMismatch(Marker::Null)) => (None, bytes.remaining_slice()), + Err(err) => return Err(error_report(err)), + }; + let (author, bytes) = Self::read_optional_string(bytes)?; + let (intent, bytes) = if nfields > 10 { + Self::read_optional_string(bytes)? + } else { + (None, bytes) + }; + + if !bytes.is_empty() { + bail!("trailing bytes in encoded history. malformed") + } + + Ok(Self { + id: id.to_owned().into(), + timestamp: OffsetDateTime::from_unix_timestamp_nanos(i128::from(timestamp))?, + duration, + exit, + command: command.to_owned(), + cwd: cwd.to_owned(), + session: session.to_owned(), + hostname: hostname.to_owned(), + author: author.unwrap_or_else(|| Self::author_from_hostname(hostname)), + intent, + deleted_at: deleted_at + .map(|t| OffsetDateTime::from_unix_timestamp_nanos(i128::from(t))) + .transpose()?, + }) + } + + fn deserialize(bytes: &[u8], version: &str) -> Result<Self> { + match version { + HISTORY_VERSION_V0 => Self::deserialize_v0(bytes), + HISTORY_VERSION_V1 => Self::deserialize_v1(bytes), + + _ => bail!("unknown version {version:?}"), + } + } +} + +#[cfg(test)] +mod tests { + use std::time::Duration; + + use time::macros::datetime; + + use crate::aclient::history::{HISTORY_VERSION, HistoryExt}; + + use super::History; + + #[test] + fn test_serialize_deserialize() { + let history = History { + id: "66d16cbee7cd47538e5c5b8b44e9006e".to_owned().into(), + timestamp: datetime!(2023-05-28 18:35:40.633872 +00:00), + duration: Duration::from_nanos(49_206_000), + exit: 0, + command: "git status".to_owned(), + cwd: "/Users/conrad.ludgate/Documents/code/atuin".to_owned(), + session: "b97d9a306f274473a203d2eba41f9457".to_owned(), + hostname: "fvfg936c0kpf:conrad.ludgate".to_owned(), + author: "conrad.ludgate".to_owned(), + intent: None, + deleted_at: None, + }; + + let serialized = history.serialize().expect("failed to serialize history"); + assert_eq!( + &serialized.0[0..3], + [205, 0, 1], + "should encode as history v1" + ); + + let deserialized = History::deserialize(&serialized.0, HISTORY_VERSION) + .expect("failed to deserialize history"); + assert_eq!(history, deserialized); + } + + #[test] + fn test_serialize_deserialize_deleted() { + let history = History { + id: "66d16cbee7cd47538e5c5b8b44e9006e".to_owned().into(), + timestamp: datetime!(2023-05-28 18:35:40.633872 +00:00), + duration: Duration::from_nanos(49_206_000), + exit: 0, + command: "git status".to_owned(), + cwd: "/Users/conrad.ludgate/Documents/code/atuin".to_owned(), + session: "b97d9a306f274473a203d2eba41f9457".to_owned(), + hostname: "fvfg936c0kpf:conrad.ludgate".to_owned(), + author: "conrad.ludgate".to_owned(), + intent: None, + deleted_at: Some(datetime!(2023-11-19 20:18 +00:00)), + }; + + let serialized = history.serialize().expect("failed to serialize history"); + + let deserialized = History::deserialize(&serialized.0, HISTORY_VERSION) + .expect("failed to deserialize history"); + + assert_eq!(history, deserialized); + } + + #[test] + fn test_serialize_deserialize_with_author_and_intent() { + let history = History { + id: "66d16cbee7cd47538e5c5b8b44e9006e".to_owned().into(), + timestamp: datetime!(2023-05-28 18:35:40.633872 +00:00), + duration: Duration::from_nanos(49_206_000), + exit: 0, + command: "git status".to_owned(), + cwd: "/Users/conrad.ludgate/Documents/code/atuin".to_owned(), + session: "b97d9a306f274473a203d2eba41f9457".to_owned(), + hostname: "fvfg936c0kpf:conrad.ludgate".to_owned(), + author: "claude".to_owned(), + intent: Some("check repository status".to_owned()), + deleted_at: None, + }; + + let serialized = history.serialize().expect("failed to serialize history"); + let deserialized = History::deserialize(&serialized.0, HISTORY_VERSION) + .expect("failed to deserialize history"); + + assert_eq!(history, deserialized); + } + + #[test] + fn test_serialize_deserialize_version() { + // v0 + let bytes_v0 = [ + 205, 0, 0, 153, 217, 32, 54, 54, 100, 49, 54, 99, 98, 101, 101, 55, 99, 100, 52, 55, + 53, 51, 56, 101, 53, 99, 53, 98, 56, 98, 52, 52, 101, 57, 48, 48, 54, 101, 207, 23, 99, + 98, 117, 24, 210, 246, 128, 206, 2, 238, 210, 240, 0, 170, 103, 105, 116, 32, 115, 116, + 97, 116, 117, 115, 217, 42, 47, 85, 115, 101, 114, 115, 47, 99, 111, 110, 114, 97, 100, + 46, 108, 117, 100, 103, 97, 116, 101, 47, 68, 111, 99, 117, 109, 101, 110, 116, 115, + 47, 99, 111, 100, 101, 47, 97, 116, 117, 105, 110, 217, 32, 98, 57, 55, 100, 57, 97, + 51, 48, 54, 102, 50, 55, 52, 52, 55, 51, 97, 50, 48, 51, 100, 50, 101, 98, 97, 52, 49, + 102, 57, 52, 53, 55, 187, 102, 118, 102, 103, 57, 51, 54, 99, 48, 107, 112, 102, 58, + 99, 111, 110, 114, 97, 100, 46, 108, 117, 100, 103, 97, 116, 101, 192, + ]; + + let deserialized = History::deserialize(&bytes_v0, "v0"); + assert!(deserialized.is_ok()); + + let deserialized = History::deserialize(&bytes_v0, HISTORY_VERSION); + assert!(deserialized.is_err()); + + let current = History { + id: "66d16cbee7cd47538e5c5b8b44e9006e".to_owned().into(), + timestamp: datetime!(2023-05-28 18:35:40.633872 +00:00), + duration: Duration::from_nanos(49_206_000), + exit: 0, + command: "git status".to_owned(), + cwd: "/Users/conrad.ludgate/Documents/code/atuin".to_owned(), + session: "b97d9a306f274473a203d2eba41f9457".to_owned(), + hostname: "fvfg936c0kpf:conrad.ludgate".to_owned(), + author: "conrad.ludgate".to_owned(), + intent: None, + deleted_at: None, + }; + + let bytes_v1 = current.serialize().expect("failed to serialize history"); + let deserialized = History::deserialize(&bytes_v1.0, HISTORY_VERSION); + assert!(deserialized.is_ok()); + + let deserialized = History::deserialize(&bytes_v1.0, "v0"); + assert!(deserialized.is_err()); + } +} diff --git a/crates/daemon/src/aclient/history/store.rs b/crates/daemon/src/aclient/history/store.rs new file mode 100644 index 00000000..a6d6a627 --- /dev/null +++ b/crates/daemon/src/aclient/history/store.rs @@ -0,0 +1,296 @@ +use eyre::{Result, bail, eyre}; +use rmp::decode::Bytes; +use turtle_api::history::{History, HistoryId}; + +use crate::aclient::{ + database::ClientSqlite, + history::HistoryExt, + record::{encryption::PASETO_V4, sqlite_store::SqliteStore}, +}; +use turtle_common::record::{DecryptedData, Host, HostId, Record, RecordId, RecordIdx}; + +use super::{HISTORY_TAG, HISTORY_VERSION, HISTORY_VERSION_V0}; + +#[derive(Debug, Clone)] +pub(crate) struct HistoryStore { + store: SqliteStore, + host_id: HostId, + encryption_key: [u8; 32], +} + +#[derive(Debug, Eq, PartialEq, Clone)] +enum HistoryRecord { + Create(History), // Create a history record + Delete(HistoryId), // Delete a history record, identified by ID +} + +impl HistoryRecord { + /// Serialize a history record, returning `DecryptedData` + /// The record will be of a certain type + /// We map those like so: + /// + /// `HistoryRecord::Create` -> 0 + /// `HistoryRecord::Delete`-> 1 + /// + /// This numeric identifier is then written as the first byte to the buffer. For history, we + /// append the serialized history right afterwards, to avoid having to handle serialization + /// twice. + /// + /// Deletion simply refers to the history by ID + fn serialize(&self) -> Result<DecryptedData> { + // probably don't actually need to use rmp here, but if we ever need to extend it, it's a + // nice wrapper around raw byte stuff + use rmp::encode; + + let mut output = vec![]; + + match self { + Self::Create(history) => { + // 0 -> a history create + encode::write_u8(&mut output, 0)?; + + let bytes = history.serialize()?; + + encode::write_bin(&mut output, &bytes.0)?; + } + Self::Delete(id) => { + // 1 -> a history delete + encode::write_u8(&mut output, 1)?; + encode::write_str(&mut output, id.to_string().as_str())?; + } + } + + Ok(DecryptedData(output)) + } + + fn deserialize(bytes: &DecryptedData, version: &str) -> Result<Self> { + use rmp::decode; + + fn error_report<E: std::fmt::Debug>(err: E) -> eyre::Report { + eyre!("{err:?}") + } + + let mut bytes = Bytes::new(&bytes.0); + + let record_type = decode::read_u8(&mut bytes).map_err(error_report)?; + + match record_type { + // 0 -> HistoryRecord::Create + 0 => { + // not super useful to us atm, but perhaps in the future + // written by write_bin above + decode::read_bin_len(&mut bytes).map_err(error_report)?; + + let record = History::deserialize(bytes.remaining_slice(), version)?; + + Ok(Self::Create(record)) + } + + // 1 -> HistoryRecord::Delete + 1 => { + let bytes = bytes.remaining_slice(); + let (id, bytes) = decode::read_str_from_slice(bytes).map_err(error_report)?; + + if !bytes.is_empty() { + bail!( + "trailing bytes decoding HistoryRecord::Delete - malformed? got {bytes:?}" + ); + } + + Ok(Self::Delete(id.to_string().into())) + } + + n => { + bail!("unknown HistoryRecord type {n}") + } + } + } +} + +impl HistoryStore { + pub(crate) fn new(store: SqliteStore, host_id: HostId, encryption_key: [u8; 32]) -> Self { + Self { + store, + host_id, + encryption_key, + } + } + + async fn push_record(&self, record: HistoryRecord) -> Result<(RecordId, RecordIdx)> { + let bytes = record.serialize()?; + let idx = self + .store + .last(self.host_id, HISTORY_TAG) + .await? + .map_or(0, |p| p.idx + 1); + + let record = Record::builder() + .host(Host::new(self.host_id)) + .version(HISTORY_VERSION.to_string()) + .tag(HISTORY_TAG.to_string()) + .idx(idx) + .data(bytes) + .build(); + + let id = record.id; + + self.store + .push(&record.encrypt::<PASETO_V4>(&self.encryption_key)) + .await?; + + Ok((id, idx)) + } + + pub(crate) async fn push(&self, history: History) -> Result<(RecordId, RecordIdx)> { + // TODO(ellie): move the history store to its own file + // it's tiny rn so fine as is + let record = HistoryRecord::Create(history); + + self.push_record(record).await + } + + // async fn build(&self, database: &ClientSqlite) -> Result<()> { + // // I'd like to change how we rebuild and not couple this with the database, but need to + // // consider the structure more deeply. This will be easy to change. + // + // // TODO(ellie): page or iterate this + // let history = self.history().await?; + // + // // In theory we could flatten this here + // // The current issue is that the database may have history in it already, from the old sync + // // This didn't actually delete old history + // // If we're sure we have a DB only maintained by the new store, we can flatten + // // create/delete before we even get to sqlite + // let mut creates = Vec::new(); + // let mut deletes = Vec::new(); + // + // for i in history { + // match i { + // HistoryRecord::Create(h) => { + // creates.push(h); + // } + // HistoryRecord::Delete(id) => { + // deletes.push(id); + // } + // } + // } + // + // database.save_bulk(&creates).await?; + // database.delete_rows(&deletes).await?; + // + // Ok(()) + // } + + pub(crate) async fn incremental_build( + &self, + database: &ClientSqlite, + ids: &[RecordId], + ) -> Result<()> { + for id in ids { + let record = self.store.get(*id).await; + + if let Ok(record) = record { + if record.tag != HISTORY_TAG { + continue; + } + + let version = record.version.clone(); + let decrypted = record.decrypt::<PASETO_V4>(&self.encryption_key)?; + let record = match version.as_str() { + HISTORY_VERSION_V0 | HISTORY_VERSION => { + HistoryRecord::deserialize(&decrypted.data, version.as_str())? + } + version => bail!("unknown history version {version:?}"), + }; + + match record { + HistoryRecord::Create(h) => { + // TODO: benchmark CPU time/memory tradeoff of batch commit vs one at a time + database.save(&h).await?; + } + HistoryRecord::Delete(id) => { + database.delete_rows(&[id]).await?; + } + } + } + } + + Ok(()) + } +} + +#[cfg(test)] +mod tests { + use std::time::Duration; + + use time::macros::datetime; + use turtle_common::record::DecryptedData; + + use crate::aclient::history::{HISTORY_VERSION, store::HistoryRecord}; + + use super::History; + + #[test] + fn test_serialize_deserialize_create() { + let bytes = [ + 204, 0, 196, 147, 205, 0, 1, 154, 217, 32, 48, 49, 56, 99, 100, 52, 102, 101, 56, 49, + 55, 53, 55, 99, 100, 50, 97, 101, 101, 54, 53, 99, 100, 55, 56, 54, 49, 102, 57, 99, + 56, 49, 207, 23, 166, 251, 212, 181, 82, 0, 0, 100, 0, 162, 108, 115, 217, 41, 47, 85, + 115, 101, 114, 115, 47, 101, 108, 108, 105, 101, 47, 115, 114, 99, 47, 103, 105, 116, + 104, 117, 98, 46, 99, 111, 109, 47, 97, 116, 117, 105, 110, 115, 104, 47, 97, 116, 117, + 105, 110, 217, 32, 48, 49, 56, 99, 100, 52, 102, 101, 97, 100, 56, 57, 55, 53, 57, 55, + 56, 53, 50, 53, 50, 55, 97, 51, 49, 99, 57, 57, 56, 48, 53, 57, 170, 98, 111, 111, 112, + 58, 101, 108, 108, 105, 101, 192, 165, 101, 108, 108, 105, 101, + ]; + + let history = History { + id: "018cd4fe81757cd2aee65cd7861f9c81".to_owned().into(), + timestamp: datetime!(2024-01-04 00:00:00.000000 +00:00), + duration: Duration::from_nanos(100), + exit: 0, + command: "ls".to_owned(), + cwd: "/Users/ellie/src/github.com/atuinsh/atuin".to_owned(), + session: "018cd4fead897597852527a31c998059".to_owned(), + hostname: "boop:ellie".to_owned(), + author: "ellie".to_owned(), + intent: None, + deleted_at: None, + }; + + let record = HistoryRecord::Create(history); + + let serialized = record.serialize().expect("failed to serialize history"); + assert_eq!(serialized.0, bytes); + + let deserialized = HistoryRecord::deserialize(&serialized, HISTORY_VERSION) + .expect("failed to deserialize HistoryRecord"); + assert_eq!(deserialized, record); + + // check the snapshot too + let deserialized = + HistoryRecord::deserialize(&DecryptedData(Vec::from(bytes)), HISTORY_VERSION) + .expect("failed to deserialize HistoryRecord"); + assert_eq!(deserialized, record); + } + + #[test] + fn test_serialize_deserialize_delete() { + let bytes = [ + 204, 1, 217, 32, 48, 49, 56, 99, 100, 52, 102, 101, 56, 49, 55, 53, 55, 99, 100, 50, + 97, 101, 101, 54, 53, 99, 100, 55, 56, 54, 49, 102, 57, 99, 56, 49, + ]; + let record = HistoryRecord::Delete("018cd4fe81757cd2aee65cd7861f9c81".to_string().into()); + + let serialized = record.serialize().expect("failed to serialize history"); + assert_eq!(serialized.0, bytes); + + let deserialized = HistoryRecord::deserialize(&serialized, HISTORY_VERSION) + .expect("failed to deserialize HistoryRecord"); + assert_eq!(deserialized, record); + + let deserialized = + HistoryRecord::deserialize(&DecryptedData(Vec::from(bytes)), HISTORY_VERSION) + .expect("failed to deserialize HistoryRecord"); + assert_eq!(deserialized, record); + } +} diff --git a/crates/daemon/src/aclient/meta.rs b/crates/daemon/src/aclient/meta.rs new file mode 100644 index 00000000..ea660745 --- /dev/null +++ b/crates/daemon/src/aclient/meta.rs @@ -0,0 +1,182 @@ +use std::path::Path; +use std::str::FromStr; +use std::time::Duration; + +use eyre::{Result, eyre}; +use sqlx::sqlite::{SqliteConnectOptions, SqliteJournalMode, SqlitePool, SqlitePoolOptions}; +use time::{OffsetDateTime, format_description::well_known::Rfc3339}; +use tokio::sync::OnceCell; +use tracing::debug; +use turtle_common::record::HostId; +use uuid::Uuid; + +const KEY_HOST_ID: &str = "host_id"; +const KEY_LAST_SYNC: &str = "last_sync_time"; + +pub(crate) struct MetaStore { + pool: SqlitePool, + cached_host_id: OnceCell<HostId>, +} + +impl MetaStore { + pub(crate) async fn new(path: impl AsRef<Path>, timeout: f64) -> Result<Self> { + let path = path.as_ref(); + let path_str = path + .as_os_str() + .to_str() + .ok_or_else(|| eyre!("meta database path is not valid UTF-8: {path:?}"))?; + debug!("opening meta sqlite database at {path:?}"); + + let is_memory = path_str.contains(":memory:"); + + if !is_memory + && !path.exists() + && let Some(dir) = path.parent() + { + fs_err::create_dir_all(dir)?; + } + + // Use DELETE journal mode instead of WAL. This is a small, infrequently- + // written KV store — WAL's concurrency benefits aren't needed, and DELETE + // mode avoids creating auxiliary -wal/-shm files that complicate + // permission handling. + let opts = SqliteConnectOptions::from_str(path_str)? + .journal_mode(SqliteJournalMode::Delete) + .optimize_on_close(true, None) + .create_if_missing(true); + + let pool = SqlitePoolOptions::new() + .acquire_timeout(Duration::from_secs_f64(timeout)) + .connect_with(opts) + .await?; + + sqlx::migrate!("./db/client-meta-migrations") + .run(&pool) + .await?; + + // Session tokens are stored in this database, so restrict permissions. + #[cfg(unix)] + if !is_memory { + use std::os::unix::fs::PermissionsExt; + std::fs::set_permissions(path, std::fs::Permissions::from_mode(0o600))?; + } + + let store = Self { + pool, + cached_host_id: OnceCell::const_new(), + }; + + Ok(store) + } + + // Generic key-value operations + + async fn get(&self, key: &str) -> Result<Option<String>> { + let row: Option<(String,)> = sqlx::query_as("SELECT value FROM meta WHERE key = ?1") + .bind(key) + .fetch_optional(&self.pool) + .await?; + + Ok(row.map(|r| r.0)) + } + + async fn set(&self, key: &str, value: &str) -> Result<()> { + sqlx::query( + " + INSERT INTO meta (key, value, updated_at) + VALUES (?1, ?2, strftime('%s', 'now')) + ON CONFLICT(key) DO UPDATE + SET value = ?2, updated_at = strftime('%s', 'now') + ", + ) + .bind(key) + .bind(value) + .execute(&self.pool) + .await?; + + Ok(()) + } + + // Typed accessors + + pub(crate) async fn host_id(&self) -> Result<HostId> { + self.cached_host_id + .get_or_try_init(|| async { + if let Some(id) = self.get(KEY_HOST_ID).await? { + let parsed = Uuid::from_str(id.as_str()) + .map_err(|e| eyre!("failed to parse host ID: {e}"))?; + return Ok(HostId(parsed)); + } + + let uuid = turtle_common::utils::uuid_v7(); + self.set(KEY_HOST_ID, uuid.as_simple().to_string().as_ref()) + .await?; + + Ok(HostId(uuid)) + }) + .await + .copied() + } + + pub(crate) async fn last_sync(&self) -> Result<OffsetDateTime> { + match self.get(KEY_LAST_SYNC).await? { + Some(v) => Ok(OffsetDateTime::parse(v.as_str(), &Rfc3339)?), + None => Ok(OffsetDateTime::UNIX_EPOCH), + } + } + + pub(crate) async fn save_sync_time(&self) -> Result<()> { + self.set( + KEY_LAST_SYNC, + OffsetDateTime::now_utc().format(&Rfc3339)?.as_str(), + ) + .await + } +} + +#[cfg(test)] +mod tests { + use super::{MetaStore, OffsetDateTime}; + + async fn new_test_store() -> MetaStore { + MetaStore::new("sqlite::memory:", 2.0).await.unwrap() + } + + #[tokio::test] + async fn test_get_set_delete() { + let store = new_test_store().await; + + assert_eq!(store.get("foo").await.unwrap(), None); + + store.set("foo", "bar").await.unwrap(); + assert_eq!(store.get("foo").await.unwrap(), Some("bar".to_string())); + + store.set("foo", "baz").await.unwrap(); + assert_eq!(store.get("foo").await.unwrap(), Some("baz".to_string())); + + // store.delete("foo").await.unwrap(); + // assert_eq!(store.get("foo").await.unwrap(), None); + } + + #[tokio::test] + async fn test_host_id_generation_and_stability() { + let store = new_test_store().await; + + let id1 = store.host_id().await.unwrap(); + let id2 = store.host_id().await.unwrap(); + + assert_eq!(id1, id2, "host_id should be stable across calls"); + } + + #[tokio::test] + async fn test_sync_time() { + let store = new_test_store().await; + + let t = store.last_sync().await.unwrap(); + assert_eq!(t, OffsetDateTime::UNIX_EPOCH); + + store.save_sync_time().await.unwrap(); + let t = store.last_sync().await.unwrap(); + assert!(t > OffsetDateTime::UNIX_EPOCH); + } +} diff --git a/crates/daemon/src/aclient/mod.rs b/crates/daemon/src/aclient/mod.rs new file mode 100644 index 00000000..2c445945 --- /dev/null +++ b/crates/daemon/src/aclient/mod.rs @@ -0,0 +1,9 @@ +pub(crate) mod database; +pub(crate) mod encryption; +pub(crate) mod history; +pub(crate) mod record; +pub(crate) mod settings; + +mod api_client; +mod meta; +mod utils; diff --git a/crates/daemon/src/aclient/record/encryption.rs b/crates/daemon/src/aclient/record/encryption.rs new file mode 100644 index 00000000..11de96d5 --- /dev/null +++ b/crates/daemon/src/aclient/record/encryption.rs @@ -0,0 +1,379 @@ +use base64::{Engine, engine::general_purpose}; +use eyre::{Context, Result, ensure}; +use rusty_paserk::{Key, KeyId, Local, PieWrappedKey}; +use rusty_paseto::core::{ + ImplicitAssertion, Key as DataKey, Local as LocalPurpose, Paseto, PasetoNonce, Payload, V4 +}; +use serde::{Deserialize, Serialize}; +use turtle_common::record::{ + AdditionalData, DecryptedData, EncryptedData, Encryption, HostId, RecordId, RecordIdx, +}; + +/// Use PASETO V4 Local encryption using the additional data as an implicit assertion. +#[expect(non_camel_case_types)] +pub(crate) struct PASETO_V4; + +/* +Why do we use a random content-encryption key? +Originally I was planning on using a derived key for encryption based on additional data. +This would be a lot more secure than using the master key directly. + +However, there's an established norm of using a random key. This scheme might be otherwise known as +- client-side encryption +- envelope encryption +- key wrapping + +A HSM (Hardware Security Module) provider, eg: AWS, Azure, GCP, or even a physical device like a YubiKey +will have some keys that they keep to themselves. These keys never leave their physical hardware. +If they never leave the hardware, then encrypting large amounts of data means giving them the data and waiting. +This is not a practical solution. Instead, generate a unique key for your data, encrypt that using your HSM +and then store that with your data. + +See + - <https://docs.aws.amazon.com/wellarchitected/latest/financial-services-industry-lens/use-envelope-encryption-with-customer-master-keys.html> + - <https://cloud.google.com/kms/docs/envelope-encryption> + - <https://learn.microsoft.com/en-us/azure/storage/blobs/client-side-encryption?tabs=dotnet#encryption-and-decryption-via-the-envelope-technique> + - <https://www.yubico.com/gb/product/yubihsm-2-fips/> + - <https://cheatsheetseries.owasp.org/cheatsheets/Cryptographic_Storage_Cheat_Sheet.html#encrypting-stored-keys> + +Why would we care? In the past we have received some requests for company solutions. If in future we can configure a +KMS service with little effort, then that would solve a lot of issues for their security team. + +Even for personal use, if a user is not comfortable with sharing keys between hosts, +GCP HSM costs $1/month and $0.03 per 10,000 key operations. Assuming an active user runs +1000 atuin records a day, that would only cost them $1 and 10 cent a month. + +Additionally, key rotations are much simpler using this scheme. Rotating a key is as simple as re-encrypting the CEK, and not the message contents. +This makes it very fast to rotate a key in bulk. + +For future reference, with asymmetric encryption, you can encrypt the CEK without the HSM's involvement, but decrypting +will need the HSM. This allows the encryption path to still be extremely fast (no network calls) but downloads/decryption +that happens in the background can make the network calls to the HSM +*/ + +impl Encryption for PASETO_V4 { + fn re_encrypt( + mut data: EncryptedData, + _ad: AdditionalData<'_>, + old_key: &[u8; 32], + new_key: &[u8; 32], + ) -> Result<EncryptedData> { + let cek = Self::decrypt_cek(&data.content_encryption_key, old_key)?; + data.content_encryption_key = Self::encrypt_cek(cek, new_key); + Ok(data) + } + + fn encrypt(data: DecryptedData, ad: AdditionalData<'_>, key: &[u8; 32]) -> EncryptedData { + // generate a random key for this entry + // aka content-encryption-key (CEK) + let random_key = Key::<V4, Local>::new_os_random(); + + let assertions = Assertions::from(ad).encode(); + + // build the payload and encrypt the token + let payload = serde_json::to_string(&AtuinPayload { + data: general_purpose::URL_SAFE_NO_PAD.encode(data.0), + }) + .expect("json encoding can't fail"); + let nonce = DataKey::<32>::try_new_random().expect("could not source from random"); + let nonce = PasetoNonce::<V4, LocalPurpose>::from(&nonce); + + let token = Paseto::<V4, LocalPurpose>::builder() + .set_payload(Payload::from(payload.as_str())) + .set_implicit_assertion(ImplicitAssertion::from(assertions.as_str())) + .try_encrypt(&random_key.into(), &nonce) + .expect("error encrypting atuin data"); + + EncryptedData { + data: token, + content_encryption_key: Self::encrypt_cek(random_key, key), + } + } + + fn decrypt( + data: EncryptedData, + ad: AdditionalData<'_>, + key: &[u8; 32], + ) -> Result<DecryptedData> { + let token = data.data; + let cek = Self::decrypt_cek(&data.content_encryption_key, key)?; + + // encode the implicit assertions + let assertions = Assertions::from(ad).encode(); + + // decrypt the payload with the footer and implicit assertions + let payload = Paseto::<V4, LocalPurpose>::try_decrypt( + &token, + &cek.into(), + None, + ImplicitAssertion::from(&*assertions), + ) + .context("could not decrypt entry")?; + + let payload: AtuinPayload = serde_json::from_str(&payload)?; + let data = general_purpose::URL_SAFE_NO_PAD.decode(payload.data)?; + Ok(DecryptedData(data)) + } +} + +impl PASETO_V4 { + fn decrypt_cek(wrapped_cek: &str, key: &[u8; 32]) -> Result<Key<V4, Local>> { + let wrapping_key = Key::<V4, Local>::from_bytes(*key); + + // let wrapping_key = PasetoSymmetricKey::from(Key::from(key)); + + let AtuinFooter { kid, wpk } = serde_json::from_str(wrapped_cek) + .context("wrapped cek did not contain the correct contents")?; + + // check that the wrapping key matches the required key to decrypt. + // In future, we could support multiple keys and use this key to + // look up the key rather than only allow one key. + // For now though we will only support the one key and key rotation will + // have to be a hard reset + let current_kid = wrapping_key.to_id(); + + ensure!( + current_kid == kid, + "attempting to decrypt with incorrect key. currently using {current_kid}, expecting {kid}" + ); + + // decrypt the random key + Ok(wpk.unwrap_key(&wrapping_key)?) + } + + fn encrypt_cek(cek: Key<V4, Local>, key: &[u8; 32]) -> String { + // aka key-encryption-key (KEK) + let wrapping_key = Key::<V4, Local>::from_bytes(*key); + let kid = wrapping_key.to_id(); + + // wrap the random key so we can decrypt it later + let wrapped_cek = AtuinFooter { + wpk: cek.wrap_pie(&wrapping_key), + kid, + }; + serde_json::to_string(&wrapped_cek).expect("could not serialize wrapped cek") + } +} + +#[derive(Serialize, Deserialize)] +struct AtuinPayload { + data: String, +} + +#[derive(Serialize, Deserialize)] +/// Well-known footer claims for decrypting. This is not encrypted but is stored in the record. +/// <https://github.com/paseto-standard/paseto-spec/blob/master/docs/02-Implementation-Guide/04-Claims.md#optional-footer-claims> +struct AtuinFooter { + /// Wrapped key + wpk: PieWrappedKey<V4, Local>, + /// ID of the key which was used to wrap + kid: KeyId<V4, Local>, +} + +/// Used in the implicit assertions. This is not encrypted and not stored in the data blob. +// This cannot be changed, otherwise it breaks the authenticated encryption. +#[derive(Debug, Copy, Clone, Serialize)] +struct Assertions<'a> { + id: &'a RecordId, + idx: &'a RecordIdx, + version: &'a str, + tag: &'a str, + host: &'a HostId, +} + +impl<'a> From<AdditionalData<'a>> for Assertions<'a> { + fn from(ad: AdditionalData<'a>) -> Self { + Self { + id: ad.id, + version: ad.version, + tag: ad.tag, + host: ad.host, + idx: ad.idx, + } + } +} + +impl Assertions<'_> { + fn encode(&self) -> String { + serde_json::to_string(self).expect("could not serialize implicit assertions") + } +} + +#[cfg(test)] +mod tests { + use turtle_common::{ + record::{Host, Record}, + utils::uuid_v7, + }; + + use super::{ + AdditionalData, DecryptedData, Encryption, HostId, Key, Local, PASETO_V4, RecordId, V4, + }; + + #[test] + fn round_trip() { + let key = Key::<V4, Local>::new_os_random(); + + let ad = AdditionalData { + id: &RecordId(uuid_v7()), + version: "v0", + tag: "kv", + host: &HostId(uuid_v7()), + idx: &0, + }; + + let data = DecryptedData(vec![1, 2, 3, 4]); + + let encrypted = PASETO_V4::encrypt(data.clone(), ad, &key.to_bytes()); + let decrypted = PASETO_V4::decrypt(encrypted, ad, &key.to_bytes()).unwrap(); + assert_eq!(decrypted, data); + } + + #[test] + fn same_entry_different_output() { + let key = Key::<V4, Local>::new_os_random(); + + let ad = AdditionalData { + id: &RecordId(uuid_v7()), + version: "v0", + tag: "kv", + host: &HostId(uuid_v7()), + idx: &0, + }; + + let data = DecryptedData(vec![1, 2, 3, 4]); + + let encrypted = PASETO_V4::encrypt(data.clone(), ad, &key.to_bytes()); + let encrypted2 = PASETO_V4::encrypt(data, ad, &key.to_bytes()); + + assert_ne!( + encrypted.data, encrypted2.data, + "re-encrypting the same contents should have different output due to key randomization" + ); + } + + #[test] + fn cannot_decrypt_different_key() { + let key = Key::<V4, Local>::new_os_random(); + let fake_key = Key::<V4, Local>::new_os_random(); + + let ad = AdditionalData { + id: &RecordId(uuid_v7()), + version: "v0", + tag: "kv", + host: &HostId(uuid_v7()), + idx: &0, + }; + + let data = DecryptedData(vec![1, 2, 3, 4]); + + let encrypted = PASETO_V4::encrypt(data, ad, &key.to_bytes()); + drop(PASETO_V4::decrypt(encrypted, ad, &fake_key.to_bytes()).unwrap_err()); + } + + #[test] + fn cannot_decrypt_different_id() { + let key = Key::<V4, Local>::new_os_random(); + + let ad = AdditionalData { + id: &RecordId(uuid_v7()), + version: "v0", + tag: "kv", + host: &HostId(uuid_v7()), + idx: &0, + }; + + let data = DecryptedData(vec![1, 2, 3, 4]); + + let encrypted = PASETO_V4::encrypt(data, ad, &key.to_bytes()); + + let ad = AdditionalData { + id: &RecordId(uuid_v7()), + ..ad + }; + drop(PASETO_V4::decrypt(encrypted, ad, &key.to_bytes()).unwrap_err()); + } + + #[test] + fn re_encrypt_round_trip() { + let key1 = Key::<V4, Local>::new_os_random(); + let key2 = Key::<V4, Local>::new_os_random(); + + let ad = AdditionalData { + id: &RecordId(uuid_v7()), + version: "v0", + tag: "kv", + host: &HostId(uuid_v7()), + idx: &0, + }; + + let data = DecryptedData(vec![1, 2, 3, 4]); + + let encrypted1 = PASETO_V4::encrypt(data.clone(), ad, &key1.to_bytes()); + let encrypted2 = + PASETO_V4::re_encrypt(encrypted1.clone(), ad, &key1.to_bytes(), &key2.to_bytes()) + .unwrap(); + + // we only re-encrypt the content keys + assert_eq!(encrypted1.data, encrypted2.data); + assert_ne!( + encrypted1.content_encryption_key, + encrypted2.content_encryption_key + ); + + let decrypted = PASETO_V4::decrypt(encrypted2, ad, &key2.to_bytes()).unwrap(); + + assert_eq!(decrypted, data); + } + + #[test] + fn full_record_round_trip() { + let key = [0x55; 32]; + let record = Record::builder() + .id(RecordId(uuid_v7())) + .version("v0".to_owned()) + .tag("kv".to_owned()) + .host(Host::new(HostId(uuid_v7()))) + .timestamp(1_687_244_806_000_000) + .data(DecryptedData(vec![1, 2, 3, 4])) + .idx(0) + .build(); + + let encrypted = record.encrypt::<PASETO_V4>(&key); + + assert!(!encrypted.data.data.is_empty()); + assert!(!encrypted.data.content_encryption_key.is_empty()); + + let decrypted = encrypted.decrypt::<PASETO_V4>(&key).unwrap(); + + assert_eq!(decrypted.data.0, [1, 2, 3, 4]); + } + + #[test] + fn full_record_round_trip_fail() { + let key = [0x55; 32]; + let record = Record::builder() + .id(RecordId(uuid_v7())) + .version("v0".to_owned()) + .tag("kv".to_owned()) + .host(Host::new(HostId(uuid_v7()))) + .timestamp(1_687_244_806_000_000) + .data(DecryptedData(vec![1, 2, 3, 4])) + .idx(0) + .build(); + + let encrypted = record.encrypt::<PASETO_V4>(&key); + + let mut enc1 = encrypted.clone(); + enc1.host = Host::new(HostId(uuid_v7())); + let _ = enc1 + .decrypt::<PASETO_V4>(&key) + .expect_err("tampering with the host should result in auth failure"); + + let mut enc2 = encrypted; + enc2.id = RecordId(uuid_v7()); + let _ = enc2 + .decrypt::<PASETO_V4>(&key) + .expect_err("tampering with the id should result in auth failure"); + } +} diff --git a/crates/daemon/src/aclient/record/mod.rs b/crates/daemon/src/aclient/record/mod.rs new file mode 100644 index 00000000..4e5774ea --- /dev/null +++ b/crates/daemon/src/aclient/record/mod.rs @@ -0,0 +1,3 @@ +pub(crate) mod encryption; +pub(crate) mod sqlite_store; +pub(crate) mod sync; diff --git a/crates/daemon/src/aclient/record/sqlite_store.rs b/crates/daemon/src/aclient/record/sqlite_store.rs new file mode 100644 index 00000000..0026690b --- /dev/null +++ b/crates/daemon/src/aclient/record/sqlite_store.rs @@ -0,0 +1,412 @@ +// Here we are using sqlite as a pretty dumb store, and will not be running any complex queries. +// Multiple stores of multiple types are all stored in one chonky table (for now), and we just index +// by tag/host + +use std::path::Path; +use std::str::FromStr; + +use eyre::{Result, eyre}; +use fs_err as fs; + +use sqlx::{ + Row, + sqlite::{SqliteConnectOptions, SqliteJournalMode, SqlitePool, SqliteRow}, +}; +use tracing::debug; + +use crate::aclient::utils::setup_db; +use turtle_common::record::{ + EncryptedData, Host, HostId, Record, RecordId, RecordIdx, RecordStatus, +}; +use turtle_common::utils; +use uuid::Uuid; + +#[derive(Debug, Clone)] +pub(crate) struct SqliteStore { + pool: SqlitePool, +} + +impl SqliteStore { + pub(crate) async fn new(path: impl AsRef<Path>, timeout: f64) -> Result<Self> { + fn mk_opts(path: &str) -> sqlx::Result<SqliteConnectOptions> { + let opts = SqliteConnectOptions::from_str(path)? + .journal_mode(SqliteJournalMode::Wal) + .foreign_keys(true) + .create_if_missing(true); + + Ok(opts) + } + + let path = path.as_ref(); + + debug!("opening sqlite database at {path:?}"); + + if utils::broken_symlink(path) { + eprintln!( + "Atuin: Sqlite db path ({}) is a broken symlink. Unable to read or create replacement.", + path.display() + ); + std::process::exit(1); + } + + if !path.exists() + && let Some(dir) = path.parent() + { + fs::create_dir_all(dir)?; + } + + let pool = setup_db!(path, timeout, mk_opts, "./db/client-record-migrations").await?; + + Ok(Self { pool }) + } + + async fn save_raw( + tx: &mut sqlx::Transaction<'_, sqlx::Sqlite>, + r: &Record<EncryptedData>, + ) -> Result<()> { + // In sqlite, we are "limited" to i64. But that is still fine, until 2262. + sqlx::query( + "insert or ignore into store(id, idx, host, tag, timestamp, version, data, cek) + values(?1, ?2, ?3, ?4, ?5, ?6, ?7, ?8)", + ) + .bind(r.id.0.as_hyphenated().to_string()) + .bind(r.idx as i64) + .bind(r.host.id.0.as_hyphenated().to_string()) + .bind(r.tag.as_str()) + .bind(r.timestamp as i64) + .bind(r.version.as_str()) + .bind(r.data.data.as_str()) + .bind(r.data.content_encryption_key.as_str()) + .execute(&mut **tx) + .await?; + + Ok(()) + } + + #[expect( + clippy::needless_pass_by_value, + reason = "this is used in a place with fixed function signature" + )] + fn query_row(row: SqliteRow) -> Record<EncryptedData> { + let idx: i64 = row.get("idx"); + let timestamp: i64 = row.get("timestamp"); + + // tbh at this point things are pretty fucked so just panic + let id = Uuid::from_str(row.get("id")).expect("invalid id UUID format in sqlite DB"); + let host = Uuid::from_str(row.get("host")).expect("invalid host UUID format in sqlite DB"); + + Record { + id: RecordId(id), + idx: idx as u64, + host: Host::new(HostId(host)), + timestamp: timestamp as u64, + tag: row.get("tag"), + version: row.get("version"), + data: EncryptedData { + data: row.get("data"), + content_encryption_key: row.get("cek"), + }, + } + } +} + +/// A record store stores records +/// In more detail - we tend to need to process this into _another_ format to actually query it. +/// As is, the record store is intended as the source of truth for arbitrary data, which could +/// be shell history, kvs, etc. +impl SqliteStore { + /// Push a record + pub(crate) async fn push(&self, record: &Record<EncryptedData>) -> Result<()> { + self.push_batch(std::iter::once(record)).await + } + + /// Push a batch of records, all in one transaction + pub(crate) async fn push_batch( + &self, + records: impl Iterator<Item = &Record<EncryptedData>> + Send + Sync, + ) -> Result<()> { + let mut tx = self.pool.begin().await?; + + for record in records { + Self::save_raw(&mut tx, record).await?; + } + + tx.commit().await?; + + Ok(()) + } + + pub(crate) async fn get(&self, id: RecordId) -> Result<Record<EncryptedData>> { + let res = sqlx::query("select * from store where store.id = ?1") + .bind(id.0.as_hyphenated().to_string()) + .map(Self::query_row) + .fetch_one(&self.pool) + .await?; + + Ok(res) + } + + pub(crate) async fn last( + &self, + host: HostId, + tag: &str, + ) -> Result<Option<Record<EncryptedData>>> { + let res = + sqlx::query("select * from store where host=?1 and tag=?2 order by idx desc limit 1") + .bind(host.0.as_hyphenated().to_string()) + .bind(tag) + .map(Self::query_row) + .fetch_one(&self.pool) + .await; + + match res { + Err(sqlx::Error::RowNotFound) => Ok(None), + Err(e) => Err(eyre!("an error occurred: {}", e)), + Ok(record) => Ok(Some(record)), + } + } + + /// Get the next `limit` records, after and including the given index + pub(crate) async fn next( + &self, + host: HostId, + tag: &str, + idx: RecordIdx, + limit: u64, + ) -> Result<Vec<Record<EncryptedData>>> { + let res = sqlx::query( + "select * from store where idx >= ?1 and host = ?2 and tag = ?3 order by idx asc limit ?4", + ) + .bind(idx as i64) + .bind(host.0.as_hyphenated().to_string()) + .bind(tag) + .bind(limit as i64) + .map(Self::query_row) + .fetch_all(&self.pool) + .await?; + + Ok(res) + } + + pub(crate) async fn status(&self) -> Result<RecordStatus> { + let mut status = RecordStatus::new(); + + let res: Result<Vec<(String, String, i64)>, sqlx::Error> = + sqlx::query_as("select host, tag, max(idx) from store group by host, tag") + .fetch_all(&self.pool) + .await; + + let res = match res { + Err(e) => return Err(eyre!("failed to fetch local store status: {}", e)), + Ok(v) => v, + }; + + for i in res { + let host = HostId( + Uuid::from_str(i.0.as_str()).expect("failed to parse uuid for local store status"), + ); + + status.set_raw(host, i.1, i.2 as u64); + } + + Ok(status) + } +} + +#[cfg(test)] +mod tests { + use crate::{ + atuin_client::{ + encryption::generate_encoded_key, record::encryption::PASETO_V4, + settings::test_local_timeout, + }, + atuin_common::{ + record::{DecryptedData, EncryptedData, Host, HostId, Record}, + utils::uuid_v7, + }, + }; + + use super::SqliteStore; + + fn test_record() -> Record<EncryptedData> { + Record::builder() + .host(Host::new(HostId(uuid_v7()))) + .version("v1".into()) + .tag(uuid_v7().simple().to_string()) + .data(EncryptedData { + data: "1234".into(), + content_encryption_key: "1234".into(), + }) + .idx(0) + .build() + } + + #[tokio::test] + async fn create_db() { + let db = SqliteStore::new(":memory:", test_local_timeout()).await; + + assert!( + db.is_ok(), + "db could not be created, {:?}", + db.err().unwrap() + ); + } + + #[tokio::test] + async fn push_record() { + let db = SqliteStore::new(":memory:", test_local_timeout()) + .await + .unwrap(); + let record = test_record(); + + db.push(&record).await.expect("failed to insert record"); + } + + #[tokio::test] + async fn get_record() { + let db = SqliteStore::new(":memory:", test_local_timeout()) + .await + .unwrap(); + let record = test_record(); + db.push(&record).await.unwrap(); + + let new_record = db.get(record.id).await.expect("failed to fetch record"); + + assert_eq!(record, new_record, "records are not equal"); + } + + #[tokio::test] + async fn last() { + let db = SqliteStore::new(":memory:", test_local_timeout()) + .await + .unwrap(); + let record = test_record(); + db.push(&record).await.unwrap(); + + let last = db + .last(record.host.id, record.tag.as_str()) + .await + .expect("failed to get store len"); + + assert_eq!( + last.unwrap().id, + record.id, + "expected to get back the same record that was inserted" + ); + } + + #[tokio::test] + async fn first() { + let db = SqliteStore::new(":memory:", test_local_timeout()) + .await + .unwrap(); + let record = test_record(); + db.push(&record).await.unwrap(); + + let first = db + .first(record.host.id, record.tag.as_str()) + .await + .expect("failed to get store len"); + + assert_eq!( + first.unwrap().id, + record.id, + "expected to get back the same record that was inserted" + ); + } + + #[tokio::test] + async fn len() { + let db = SqliteStore::new(":memory:", test_local_timeout()) + .await + .unwrap(); + let record = test_record(); + db.push(&record).await.unwrap(); + + let len = db + .len(record.host.id, record.tag.as_str()) + .await + .expect("failed to get store len"); + + assert_eq!(len, 1, "expected length of 1 after insert"); + } + + #[tokio::test] + async fn len_tag() { + let db = SqliteStore::new(":memory:", test_local_timeout()) + .await + .unwrap(); + let record = test_record(); + db.push(&record).await.unwrap(); + + let len = db + .len_tag(record.tag.as_str()) + .await + .expect("failed to get store len"); + + assert_eq!(len, 1, "expected length of 1 after insert"); + } + + #[tokio::test] + async fn re_encrypt() { + let store = SqliteStore::new(":memory:", test_local_timeout()) + .await + .unwrap(); + let (key, _) = generate_encoded_key().unwrap(); + let data = vec![0u8, 1u8, 2u8, 3u8]; + let host_id = HostId(uuid_v7()); + + for i in 0..10 { + let record = Record::builder() + .host(Host::new(host_id)) + .version(String::from("test")) + .tag(String::from("test")) + .idx(i) + .data(DecryptedData(data.clone())) + .build(); + + let record = record.encrypt::<PASETO_V4>(&key.into()); + store + .push(&record) + .await + .expect("failed to push encrypted record"); + } + + // first, check that we can decrypt the data with the current key + let all = store.all_tagged("test").await.unwrap(); + + assert_eq!(all.len(), 10, "failed to fetch all records"); + + for record in all { + let decrypted = record.decrypt::<PASETO_V4>(&key.into()).unwrap(); + assert_eq!(decrypted.data.0, data); + } + + // reencrypt the store, then check if + // 1) it cannot be decrypted with the old key + // 2) it can be decrypted with the new key + + let (new_key, _) = generate_encoded_key().unwrap(); + store + .re_encrypt(&key.into(), &new_key.into()) + .await + .expect("failed to re-encrypt store"); + + let all = store.all_tagged("test").await.unwrap(); + + for record in all.iter() { + let decrypted = record.clone().decrypt::<PASETO_V4>(&key.into()); + assert!( + decrypted.is_err(), + "did not get error decrypting with old key after re-encrypt" + ) + } + + for record in all { + let decrypted = record.decrypt::<PASETO_V4>(&new_key.into()).unwrap(); + assert_eq!(decrypted.data.0, data); + } + + assert_eq!(store.len(host_id, "test").await.unwrap(), 10); + } +} diff --git a/crates/daemon/src/aclient/record/sync.rs b/crates/daemon/src/aclient/record/sync.rs new file mode 100644 index 00000000..79239b99 --- /dev/null +++ b/crates/daemon/src/aclient/record/sync.rs @@ -0,0 +1,453 @@ +// do a sync :O +use std::{cmp::Ordering, fmt::Write}; + +use eyre::{OptionExt, Result}; +use thiserror::Error; +use tracing::error; + +use super::encryption::PASETO_V4; +use crate::aclient::record::sqlite_store::SqliteStore; +use crate::aclient::{api_client::Client, settings::Settings}; + +use indicatif::{ProgressBar, ProgressState, ProgressStyle}; +use turtle_common::record::{Diff, HostId, RecordId, RecordIdx, RecordStatus}; + +#[derive(Error, Debug)] +pub(crate) enum SyncError { + #[error("an issue with the local database occurred: {msg:?}")] + LocalStoreError { msg: String }, + + #[error("something has gone wrong with the sync logic: {msg:?}")] + SyncLogicError { msg: String }, + + #[error("operational error: {msg:?}")] + OperationalError { msg: String }, + + #[error("a request to the sync server failed: {msg}")] + RemoteRequestError { msg: String }, + + #[error( + "the encryption key on this machine does not match the data on the server. \ + this usually means a new machine was set up without copying the existing key. \ + to fix: run `atuin key` on a machine that already syncs correctly, then run \ + `atuin store rekey <key>` on this machine with the value from the other machine" + )] + WrongKey, +} + +#[derive(Debug, Eq, PartialEq)] +enum Operation { + // Either upload or download until the states matches the below + Upload { + local: RecordIdx, + remote: Option<RecordIdx>, + host: HostId, + tag: String, + }, + Download { + local: Option<RecordIdx>, + remote: RecordIdx, + host: HostId, + tag: String, + }, + Noop { + host: HostId, + tag: String, + }, +} + +fn build_client(settings: &Settings) -> Result<Client<'_>, SyncError> { + Client::new( + &settings.sync.address, + settings.network_connect_timeout, + settings.network_timeout, + settings + .sync + .user_id() + .map_err(|e| SyncError::RemoteRequestError { msg: e.to_string() })? + .ok_or_eyre("No sync user-id set") + .map_err(|e| SyncError::RemoteRequestError { msg: e.to_string() })?, + ) + .map_err(|e| SyncError::OperationalError { msg: e.to_string() }) +} + +async fn diff( + client: &Client<'_>, + store: &SqliteStore, +) -> Result<(Vec<Diff>, RecordStatus), SyncError> { + let local_index = store + .status() + .await + .map_err(|e| SyncError::LocalStoreError { msg: e.to_string() })?; + + let remote_index = client + .record_status() + .await + .map_err(|e| SyncError::RemoteRequestError { msg: e.to_string() })?; + + let diff = local_index.diff(&remote_index); + + Ok((diff, remote_index)) +} + +// Take a diff, along with a local store, and resolve it into a set of operations. +// With the store as context, we can determine if a tail exists locally or not and therefore if it needs uploading or download. +// In theory this could be done as a part of the diffing stage, but it's easier to reason +// about and test this way +fn operations(diffs: Vec<Diff>, _store: &SqliteStore) -> Result<Vec<Operation>, SyncError> { + let mut operations = Vec::with_capacity(diffs.len()); + + for diff in diffs { + let op = match (diff.local, diff.remote) { + // We both have it! Could be either. Compare. + (Some(local), Some(remote)) => match local.cmp(&remote) { + Ordering::Equal => Operation::Noop { + host: diff.host, + tag: diff.tag, + }, + Ordering::Greater => Operation::Upload { + local, + remote: Some(remote), + host: diff.host, + tag: diff.tag, + }, + Ordering::Less => Operation::Download { + local: Some(local), + remote, + host: diff.host, + tag: diff.tag, + }, + }, + + // Remote has it, we don't. Gotta be download + (None, Some(remote)) => Operation::Download { + local: None, + remote, + host: diff.host, + tag: diff.tag, + }, + + // We have it, remote doesn't. Gotta be upload. + (Some(local), None) => Operation::Upload { + local, + remote: None, + host: diff.host, + tag: diff.tag, + }, + + // something is pretty fucked. + (None, None) => { + return Err(SyncError::SyncLogicError { + msg: String::from( + "diff has nothing for local or remote - (host, tag) does not exist", + ), + }); + } + }; + + operations.push(op); + } + + // sort them - purely so we have a stable testing order, and can rely on + // same input = same output + // We can sort by ID so long as we continue to use UUIDv7 or something + // with the same properties + + operations.sort_by_key(|op| match op { + Operation::Noop { host, tag } => (0, *host, tag.clone()), + + Operation::Upload { host, tag, .. } => (1, *host, tag.clone()), + + Operation::Download { host, tag, .. } => (2, *host, tag.clone()), + }); + + Ok(operations) +} + +async fn sync_upload( + store: &SqliteStore, + client: &Client<'_>, + host: HostId, + tag: String, + local: RecordIdx, + remote: Option<RecordIdx>, + page_size: u64, +) -> Result<i64, SyncError> { + let remote = remote.unwrap_or(0); + let expected = local - remote; + let mut progress = 0; + + let pb = ProgressBar::new(expected); + pb.set_style(ProgressStyle::with_template("{spinner:.green} [{elapsed_precise}] [{wide_bar:.cyan/blue}] {human_pos}/{human_len} ({eta})") + .unwrap() + .with_key("eta", |state: &ProgressState, w: &mut dyn Write| write!(w, "{:.1}s", state.eta().as_secs_f64()).unwrap()) + .progress_chars("#>-")); + + println!( + "Uploading {} records to {}/{}", + expected, + host.0.as_simple(), + tag + ); + + loop { + let page = store + .next(host, tag.as_str(), remote + progress, page_size) + .await + .map_err(|e| { + error!("failed to read upload page: {e:?}"); + + SyncError::LocalStoreError { msg: e.to_string() } + })?; + + if page.is_empty() { + break; + } + + client.post_records(&page).await.map_err(|e| { + error!("failed to post records: {e:?}"); + + SyncError::RemoteRequestError { msg: e.to_string() } + })?; + + progress += page.len() as u64; + pb.set_position(progress); + + if progress >= expected { + break; + } + } + + pb.finish_with_message("Uploaded records"); + + Ok(progress as i64) +} + +async fn sync_download( + store: &SqliteStore, + client: &Client<'_>, + host: HostId, + tag: String, + local: Option<RecordIdx>, + remote: RecordIdx, + page_size: u64, +) -> Result<Vec<RecordId>, SyncError> { + let local = local.unwrap_or(0); + let expected = remote - local; + let mut progress = 0; + let mut ret = Vec::new(); + + println!( + "Downloading {} records from {}/{}", + expected, + host.0.as_simple(), + tag + ); + + let pb = ProgressBar::new(expected); + pb.set_style(ProgressStyle::with_template("{spinner:.green} [{elapsed_precise}] [{wide_bar:.cyan/blue}] {human_pos}/{human_len} ({eta})") + .unwrap() + .with_key("eta", |state: &ProgressState, w: &mut dyn Write| write!(w, "{:.1}s", state.eta().as_secs_f64()).unwrap()) + .progress_chars("#>-")); + + loop { + let page = client + .next_records(host, tag.clone(), local + progress, page_size) + .await + .map_err(|e| SyncError::RemoteRequestError { msg: e.to_string() })?; + + if page.is_empty() { + break; + } + + store + .push_batch(page.iter()) + .await + .map_err(|e| SyncError::LocalStoreError { msg: e.to_string() })?; + + ret.extend(page.iter().map(|f| f.id)); + + progress += page.len() as u64; + pb.set_position(progress); + + if progress >= expected { + break; + } + } + + pb.finish_with_message("Downloaded records"); + + Ok(ret) +} + +async fn sync_remote( + client: &Client<'_>, + operations: Vec<Operation>, + local_store: &SqliteStore, + page_size: u64, +) -> Result<(i64, Vec<RecordId>), SyncError> { + let mut uploaded = 0; + let mut downloaded = Vec::new(); + + // this can totally run in parallel, but lets get it working first + for i in operations { + match i { + Operation::Upload { + host, + tag, + local, + remote, + } => { + uploaded += + sync_upload(local_store, client, host, tag, local, remote, page_size).await?; + } + + Operation::Download { + host, + tag, + local, + remote, + } => { + let mut d = + sync_download(local_store, client, host, tag, local, remote, page_size).await?; + downloaded.append(&mut d); + } + + Operation::Noop { .. } => (), + } + } + + Ok((uploaded, downloaded)) +} + +async fn check_encryption_key( + client: &Client<'_>, + remote_index: &RecordStatus, + encryption_key: &[u8; 32], +) -> Result<(), SyncError> { + let sample = remote_index + .hosts + .iter() + .flat_map(|(host, tags)| tags.keys().map(move |tag| (*host, tag.clone()))) + .next(); + + let Some((host, tag)) = sample else { + return Ok(()); + }; + + let records = client + .next_records(host, tag, 0, 1) + .await + .map_err(|e| SyncError::RemoteRequestError { msg: e.to_string() })?; + + let Some(record) = records.into_iter().next() else { + return Ok(()); + }; + + record.decrypt::<PASETO_V4>(encryption_key).map_err(|err| { + error!("Wrong key error: {err}"); + SyncError::WrongKey + })?; + + Ok(()) +} + +pub(crate) async fn sync( + settings: &Settings, + store: &SqliteStore, + encryption_key: &[u8; 32], +) -> Result<(i64, Vec<RecordId>), SyncError> { + let client = build_client(settings)?; + let (diff, remote_index) = diff(&client, store).await?; + + // Bail before mutating either side if the local key can't read the remote. + check_encryption_key(&client, &remote_index, encryption_key).await?; + + let operations = operations(diff, store)?; + let (uploaded, downloaded) = sync_remote(&client, operations, store, 100).await?; + + Ok((uploaded, downloaded)) +} + +#[cfg(test)] +mod tests { + use crate::aclient::record::sync::Operation; + use turtle_common::record::{Diff, EncryptedData, HostId, Record}; + + use crate::aclient::{ + record::{ + sqlite_store::SqliteStore, + sync::{self}, + }, + settings::test_local_timeout, + }; + + fn test_record() -> Record<EncryptedData> { + Record::builder() + .host(turtle_common::record::Host::new(HostId( + turtle_common::utils::uuid_v7(), + ))) + .version("v1".into()) + .tag(turtle_common::utils::uuid_v7().simple().to_string()) + .data(EncryptedData { + data: String::new(), + content_encryption_key: String::new(), + }) + .idx(0) + .build() + } + + // Take a list of local records, and a list of remote records. + // Return the local database, and a diff of local/remote, ready to build + // ops + async fn build_test_diff( + local_records: Vec<Record<EncryptedData>>, + remote_records: Vec<Record<EncryptedData>>, + ) -> (SqliteStore, Vec<Diff>) { + let local_store = SqliteStore::new(":memory:", test_local_timeout()) + .await + .expect("failed to open in memory sqlite"); + let remote_store = SqliteStore::new(":memory:", test_local_timeout()) + .await + .expect("failed to open in memory sqlite"); // "remote" + + for i in local_records { + local_store.push(&i).await.unwrap(); + } + + for i in remote_records { + remote_store.push(&i).await.unwrap(); + } + + let local_index = local_store.status().await.unwrap(); + let remote_index = remote_store.status().await.unwrap(); + + let diff = local_index.diff(&remote_index); + + (local_store, diff) + } + + #[tokio::test] + async fn test_basic_diff() { + // a diff where local is ahead of remote. nothing else. + + let record = test_record(); + let (store, diff) = build_test_diff(vec![record.clone()], vec![]).await; + + assert_eq!(diff.len(), 1); + + let operations = sync::operations(diff, &store).unwrap(); + + assert_eq!(operations.len(), 1); + + assert_eq!( + operations[0], + Operation::Upload { + host: record.host.id, + tag: record.tag, + local: record.idx, + remote: None, + } + ); + } +} diff --git a/crates/daemon/src/aclient/settings/meta.rs b/crates/daemon/src/aclient/settings/meta.rs new file mode 100644 index 00000000..1c9b9cd1 --- /dev/null +++ b/crates/daemon/src/aclient/settings/meta.rs @@ -0,0 +1,17 @@ +use serde::{Deserialize, Serialize}; + +#[derive(Debug, Serialize, Deserialize, Clone)] +pub(crate) struct Settings { + pub(super) db_path: String, +} + +impl Default for Settings { + fn default() -> Self { + let dir = turtle_common::utils::data_dir(); + let path = dir.join("meta.db"); + + Self { + db_path: path.to_string_lossy().to_string(), + } + } +} diff --git a/crates/daemon/src/aclient/settings/mod.rs b/crates/daemon/src/aclient/settings/mod.rs new file mode 100644 index 00000000..379ee563 --- /dev/null +++ b/crates/daemon/src/aclient/settings/mod.rs @@ -0,0 +1,590 @@ +use crypto_secretbox::Key; +use std::{collections::HashMap, fs::read_to_string, path::PathBuf, sync::OnceLock}; +use tokio::sync::OnceCell; +use tracing::info; +use uuid::Uuid; + +use crate::aclient::encryption::decode_key; +use clap::ValueEnum; +use config::{ + Config, ConfigBuilder, Environment, File as ConfigFile, FileFormat, builder::DefaultState, +}; +use eyre::{Context, Result, eyre}; +use fs_err::create_dir_all; +use serde::{Deserialize, Serialize}; +use time::OffsetDateTime; +use turtle_common::record::HostId; +use turtle_common::utils; + +static DATA_DIR: OnceLock<PathBuf> = OnceLock::new(); +static META_CONFIG: OnceLock<(String, f64)> = OnceLock::new(); +static META_STORE: OnceCell<crate::aclient::meta::MetaStore> = OnceCell::const_new(); + +mod meta; + +// FIXME: Can use upstream Dialect enum if https://github.com/stevedonovan/chrono-english/pull/16 is merged +// FIXME: Above PR was merged, but dependency was changed to interim (fork of chrono-english) in the ... interim +#[derive(Clone, Debug, Deserialize, Copy, Serialize)] +enum Dialect { + #[serde(rename = "us")] + Us, + + #[serde(rename = "uk")] + Uk, +} + +impl From<Dialect> for interim::Dialect { + fn from(d: Dialect) -> Self { + match d { + Dialect::Uk => Self::Uk, + Dialect::Us => Self::Us, + } + } +} + +#[derive(Clone, Debug, Deserialize, Copy, PartialEq, Eq, ValueEnum, Serialize)] +enum KeymapMode { + #[serde(rename = "emacs")] + Emacs, + + #[serde(rename = "vim-normal")] + VimNormal, + + #[serde(rename = "vim-insert")] + VimInsert, + + #[serde(rename = "auto")] + Auto, +} + +// We want to translate the config to crossterm::cursor::SetCursorStyle, but +// the original type does not implement trait serde::Deserialize unfortunately. +// It seems impossible to implement Deserialize for external types when it is +// used in HashMap (https://stackoverflow.com/questions/67142663). We instead +// define an adapter type. +#[derive(Clone, Debug, Deserialize, Copy, PartialEq, Eq, ValueEnum, Serialize)] +enum CursorStyle { + #[serde(rename = "default")] + DefaultUserShape, + + #[serde(rename = "blink-block")] + BlinkingBlock, + + #[serde(rename = "steady-block")] + SteadyBlock, + + #[serde(rename = "blink-underline")] + BlinkingUnderScore, + + #[serde(rename = "steady-underline")] + SteadyUnderScore, + + #[serde(rename = "blink-bar")] + BlinkingBar, + + #[serde(rename = "steady-bar")] + SteadyBar, +} + +#[derive(Clone, Debug, Deserialize, Serialize)] +pub(crate) struct Daemon { + /// The daemon will handle sync on an interval. How often to sync, in seconds. + pub(crate) sync_frequency: u64, + + /// The path to the unix socket used by the daemon + pub(crate) socket_path: String, + + /// Path to the daemon pidfile used for process coordination. + pub(crate) pidfile_path: String, + + /// Use a socket passed via systemd's socket activation protocol, instead of the path + pub(crate) systemd_socket: bool, + + /// The port that should be used for TCP on non unix systems + tcp_port: u64, +} + +impl Default for Daemon { + fn default() -> Self { + Self { + sync_frequency: 300, + socket_path: String::new(), + pidfile_path: String::new(), + systemd_socket: false, + tcp_port: 8889, + } + } +} + +// The preview height strategy also takes max_preview_height into account. +#[derive(Clone, Debug, Deserialize, Copy, PartialEq, Eq, ValueEnum, Serialize)] +enum PreviewStrategy { + // Preview height is calculated for the length of the selected command. + #[serde(rename = "auto")] + Auto, + + // Preview height is calculated for the length of the longest command stored in the history. + #[serde(rename = "static")] + Static, + + // max_preview_height is used as fixed height. + #[serde(rename = "fixed")] + Fixed, +} + +/// Sync-specific settings. +#[derive(Clone, Debug, Default, Deserialize, Serialize)] +pub(crate) struct Sync { + /// The sync address for atuin. + pub(crate) address: String, + + #[serde(default)] + frequency: String, + + #[serde(default)] + pub(crate) auto: bool, + + #[serde(default)] + user_id_path: Option<PathBuf>, + + #[serde(default)] + pub(crate) encryption_key_path: Option<PathBuf>, +} + +impl Sync { + fn try_read_file(file: Option<&PathBuf>) -> Result<Option<String>> { + if let Some(path) = file { + if path.try_exists()? { + let user = read_to_string(path)?; + + if user.is_empty() { + Ok(None) + } else { + Ok(Some(user)) + } + } else { + // It's okay that the file doesn't exist. + // The important part is to error out if we can't access it (e.g. Because of missing + // permissions). + Ok(None) + } + } else { + Ok(None) + } + } + + pub(crate) fn have_sync_user(&self) -> Result<bool> { + let sa = self.user_id()?; + Ok(sa.is_some()) + } + + pub(crate) fn user_id(&self) -> Result<Option<Uuid>> { + Self::try_read_file(self.user_id_path.as_ref())? + .map(|file| { + Uuid::parse_str(file.trim()).context( + "Failed to decode user id as UUID, while trying to decode sync user_id", + ) + }) + .transpose() + } + pub(crate) fn encryption_key(&self) -> Result<Option<Key>> { + Self::try_read_file(self.encryption_key_path.as_ref())? + .as_deref() + .map(str::trim) + .map(decode_key) + .transpose() + } +} + +#[derive(Clone, Debug, Deserialize, Serialize)] +pub(crate) struct Settings { + pub(crate) db_path: String, + pub(crate) record_store_path: String, + + pub(crate) network_connect_timeout: u64, + pub(crate) network_timeout: u64, + pub(crate) local_timeout: f64, + + #[serde(default)] + pub(crate) sync: Sync, + + #[serde(default)] + pub(crate) daemon: Daemon, + + #[serde(default)] + meta: meta::Settings, +} + +impl Settings { + // -- Meta store: lazily initialized on first access -- + + async fn meta_store() -> Result<&'static crate::aclient::meta::MetaStore> { + META_STORE + .get_or_try_init(|| async { + let (db_path, timeout) = META_CONFIG.get().ok_or_else(|| { + eyre!("meta store config not set — Settings::new() has not been called") + })?; + crate::aclient::meta::MetaStore::new(db_path, *timeout).await + }) + .await + } + + pub(crate) async fn host_id() -> Result<HostId> { + Self::meta_store().await?.host_id().await + } + + async fn last_sync() -> Result<OffsetDateTime> { + Self::meta_store().await?.last_sync().await + } + + pub(crate) async fn save_sync_time() -> Result<()> { + Self::meta_store().await?.save_sync_time().await + } + + fn builder() -> Result<ConfigBuilder<DefaultState>> { + Self::builder_with_data_dir(&utils::data_dir()) + } + + #[expect(clippy::too_many_lines)] + fn builder_with_data_dir(data_dir: &std::path::Path) -> Result<ConfigBuilder<DefaultState>> { + let db_path = data_dir.join("history.db"); + let record_store_path = data_dir.join("records.db"); + let kv_path = data_dir.join("kv.db"); + let scripts_path = data_dir.join("scripts.db"); + let ai_sessions_path = data_dir.join("ai_sessions.db"); + let socket_path = utils::daemon_socket_path(); + let pidfile_path = data_dir.join("atuin-daemon.pid"); + + let key_path = data_dir.join("key"); + let meta_path = data_dir.join("meta.db"); + + Ok(Config::builder() + .set_default("history_format", "{time}\t{command}\t{duration}")? + .set_default("db_path", db_path.to_str())? + .set_default("record_store_path", record_store_path.to_str())? + .set_default("key_path", key_path.to_str())? + .set_default("dialect", "us")? + .set_default("timezone", "local")? + .set_default("auto_sync", true)? + .set_default("sync.address", "https://api.atuin.sh")? + .set_default("sync_frequency", "5m")? + .set_default("search_mode", "fuzzy")? + .set_default("filter_mode", None::<String>)? + .set_default("style", "compact")? + .set_default("inline_height", 40)? + .set_default("show_preview", true)? + .set_default("preview.strategy", "auto")? + .set_default("max_preview_height", 4)? + .set_default("show_help", true)? + .set_default("show_tabs", true)? + .set_default("show_numeric_shortcuts", true)? + .set_default("auto_hide_height", 8)? + .set_default("invert", false)? + .set_default("exit_mode", "return-original")? + .set_default("word_jump_mode", "emacs")? + .set_default( + "word_chars", + "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789", + )? + .set_default("scroll_context_lines", 1)? + .set_default("shell_up_key_binding", false)? + .set_default("workspaces", false)? + .set_default("ctrl_n_shortcuts", false)? + .set_default("secrets_filter", true)? + .set_default("strip_trailing_whitespace", true)? + .set_default("network_connect_timeout", 5)? + .set_default("network_timeout", 30)? + .set_default("local_timeout", 2.0)? + // enter_accept defaults to false here, but true in the default config file. The dissonance is + // intentional! + // Existing users will get the default "False", so we don't mess with any potential + // muscle memory. + // New users will get the new default, that is more similar to what they are used to. + .set_default("enter_accept", false)? + .set_default("keys.scroll_exits", true)? + .set_default("keys.accept_past_line_end", true)? + .set_default("keys.exit_past_line_start", true)? + .set_default("keys.accept_past_line_start", false)? + .set_default("keys.accept_with_backspace", false)? + .set_default("keys.prefix", "a")? + .set_default("keymap_mode", "emacs")? + .set_default("keymap_mode_shell", "auto")? + .set_default("keymap_cursor", HashMap::<String, String>::new())? + .set_default("smart_sort", false)? + .set_default("command_chaining", false)? + .set_default("store_failed", true)? + .set_default("daemon.sync_frequency", 300)? + .set_default("daemon.socket_path", socket_path.to_str())? + .set_default("daemon.pidfile_path", pidfile_path.to_str())? + .set_default("daemon.systemd_socket", false)? + .set_default("daemon.tcp_port", 8889)? + .set_default("logs.enabled", true)? + .set_default("logs.level", "info")? + .set_default("logs.search.file", "search.log")? + .set_default("logs.daemon.file", "daemon.log")? + .set_default("logs.ai.file", "ai.log")? + .set_default("kv.db_path", kv_path.to_str())? + .set_default("scripts.db_path", scripts_path.to_str())? + .set_default("search.recency_score_multiplier", 1.0)? + .set_default("search.frequency_score_multiplier", 1.0)? + .set_default("search.frecency_score_multiplier", 1.0)? + .set_default("meta.db_path", meta_path.to_str())? + .set_default("ai.db_path", ai_sessions_path.to_str())? + .set_default("ai.session_continue_minutes", 60)? + .set_default("ai.send_cwd", false)? + .set_default("ai.opening.send_cwd", false)? + .set_default("ai.opening.send_last_command", false)? + .set_default( + "search.filters", + vec![ + "global", + "host", + "session", + "workspace", + "directory", + "session-preload", + ], + )? + .set_default("theme.name", "default")? + .set_default("theme.debug", None::<bool>)? + .set_default("tmux.enabled", false)? + .set_default("tmux.width", "80%")? + .set_default("tmux.height", "60%")? + .set_default( + "prefers_reduced_motion", + std::env::var("NO_MOTION").ok().map_or_else( + || config::Value::new(None, config::ValueKind::Boolean(false)), + |_| config::Value::new(None, config::ValueKind::Boolean(true)), + ), + )? + .set_default("no_mouse", false)? + .add_source( + Environment::with_prefix("atuin") + .prefix_separator("_") + .separator("__"), + )) + } + + pub(crate) fn get_config_path() -> Result<PathBuf> { + let config_dir = utils::config_dir(); + + create_dir_all(&config_dir) + .wrap_err_with(|| format!("could not create dir {}", config_dir.display()))?; + + let mut config_file = std::env::var("ATUIN_CONFIG_DIR").map_or_else( + |_| { + let mut config_file = PathBuf::new(); + config_file.push(config_dir); + config_file + }, + PathBuf::from, + ); + + config_file.push("config.toml"); + + Ok(config_file) + } + + /// Build a merged `Config` from defaults, config file, and environment. + /// + /// This resolves `data_dir`, initializes the data directory on disk, + /// and layers defaults → config file → env overrides. Both `new()` and + /// `get_config_value()` use this so the resolution logic lives in one place. + fn build_config() -> Result<Config> { + let config_file = Self::get_config_path()?; + + // extract data_dir first so we can use it as the base for other path defaults + let effective_data_dir = if config_file.exists() { + #[derive(Deserialize, Default)] + struct DataDirOnly { + data_dir: Option<String>, + } + + let config_file_str = config_file + .to_str() + .ok_or_else(|| eyre!("config file path is not valid UTF-8"))?; + + let partial_config = Config::builder() + .add_source(ConfigFile::new(config_file_str, FileFormat::Toml)) + .add_source( + Environment::with_prefix("atuin") + .prefix_separator("_") + .separator("__"), + ) + .build() + .ok(); + + let custom_data_dir = partial_config + .and_then(|c| c.try_deserialize::<DataDirOnly>().ok()) + .and_then(|d| d.data_dir); + + match custom_data_dir { + Some(dir) => { + let expanded = shellexpand::full(&dir) + .map_err(|e| eyre!("failed to expand data_dir path: {}", e))?; + PathBuf::from(expanded.as_ref()) + } + None => utils::data_dir(), + } + } else { + utils::data_dir() + }; + + DATA_DIR.set(effective_data_dir.clone()).ok(); + + create_dir_all(&effective_data_dir) + .wrap_err_with(|| format!("could not create dir {}", effective_data_dir.display()))?; + + let mut config_builder = Self::builder_with_data_dir(&effective_data_dir)?; + + config_builder = if config_file.exists() { + let config_file_str = config_file + .to_str() + .ok_or_else(|| eyre!("config file path is not valid UTF-8"))?; + config_builder.add_source(ConfigFile::new(config_file_str, FileFormat::Toml)) + } else { + // TODO(@bpeetz): Rework the config handling, so that we can actually auto-write a + // file with defaults. <2026-06-13> + create_dir_all(config_file.parent().unwrap())?; + + info!( + "No config file at: `{}`. Not adding one.", + config_file.display() + ); + + config_builder + }; + + // all paths should be expanded + let built = config_builder.build_cloned()?; + config_builder = [ + "db_path", + "record_store_path", + "key_path", + "daemon.socket_path", + "daemon.pidfile_path", + "logs.dir", + "logs.search.file", + "logs.daemon.file", + ] + .iter() + .map(|key| (key, built.get_string(key).unwrap_or_default())) + .filter_map(|(key, value)| match Self::expand_path(&value) { + Ok(expanded) => Some((key, expanded)), + Err(e) => { + log::warn!("failed to expand path for {key}: {e}"); + None + } + }) + .fold(config_builder, |builder, (key, value)| { + builder + .set_override(key, value) + .unwrap_or_else(|_| panic!("failed to set absolute path override for {key}")) + }); + + config_builder.build().map_err(Into::into) + } + + pub(crate) fn new() -> Result<Self> { + let config = Self::build_config()?; + let settings: Self = config + .try_deserialize() + .map_err(|e| eyre!("failed to deserialize: {}", e))?; + + // Register meta store config for lazy initialization on first access + META_CONFIG + .set((settings.meta.db_path.clone(), settings.local_timeout)) + .ok(); + + Ok(settings) + } + + fn expand_path(path: &str) -> Result<String> { + shellexpand::full(&path) + .map(|p| p.to_string()) + .map_err(|e| eyre!("failed to expand path: {}", e)) + } + + pub(crate) fn paths_ok(&self) -> bool { + let mut paths: Vec<&str> = vec![ + &self.db_path, + &self.record_store_path, + &self.meta.db_path, + &self.daemon.socket_path, + ]; + + if let Some(path) = &self.sync.encryption_key_path { + paths.push(path.to_str().unwrap()); + } + if let Some(path) = &self.sync.user_id_path { + paths.push(path.to_str().unwrap()); + } + + paths.iter().all(|p| !utils::broken_symlink(p)) + } +} + +impl Default for Settings { + fn default() -> Self { + // if this panics something is very wrong, as the default config + // does not build or deserialize into the settings struct + Self::builder() + .expect("Could not build default") + .build() + .expect("Could not build config") + .try_deserialize() + .expect("Could not deserialize config") + } +} + +#[cfg(test)] +pub(crate) fn test_local_timeout() -> f64 { + std::env::var("ATUIN_TEST_LOCAL_TIMEOUT") + .ok() + .and_then(|x| x.parse().ok()) + // this hardcoded value should be replaced by a simple way to get the + // default local_timeout of Settings if possible + .unwrap_or(2.0) +} + +#[cfg(test)] +mod tests { + use eyre::Result; + + #[test] + fn builder_with_data_dir_uses_custom_paths() -> Result<()> { + use std::path::PathBuf; + + let custom_dir = PathBuf::from("/custom/data/dir"); + let builder = super::Settings::builder_with_data_dir(&custom_dir)?; + let config = builder.build()?; + + let db_path: String = config.get("db_path")?; + let key_path: String = config.get("key_path")?; + let record_store_path: String = config.get("record_store_path")?; + let kv_db_path: String = config.get("kv.db_path")?; + let scripts_db_path: String = config.get("scripts.db_path")?; + let meta_db_path: String = config.get("meta.db_path")?; + let daemon_socket_path: String = config.get("daemon.socket_path")?; + let daemon_pidfile_path: String = config.get("daemon.pidfile_path")?; + + assert_eq!(db_path, custom_dir.join("history.db").to_str().unwrap()); + assert_eq!(key_path, custom_dir.join("key").to_str().unwrap()); + assert_eq!( + record_store_path, + custom_dir.join("records.db").to_str().unwrap() + ); + assert_eq!(kv_db_path, custom_dir.join("kv.db").to_str().unwrap()); + assert_eq!( + scripts_db_path, + custom_dir.join("scripts.db").to_str().unwrap() + ); + assert_eq!(meta_db_path, custom_dir.join("meta.db").to_str().unwrap()); + assert_eq!( + daemon_pidfile_path, + custom_dir.join("atuin-daemon.pid").to_str().unwrap() + ); + + Ok(()) + } +} diff --git a/crates/daemon/src/aclient/utils.rs b/crates/daemon/src/aclient/utils.rs new file mode 100644 index 00000000..18e732a0 --- /dev/null +++ b/crates/daemon/src/aclient/utils.rs @@ -0,0 +1,77 @@ +/// Setup a [`SQLite`] database. +/// +/// This takes care of correct locking, so that we avoid a race when setting up the database. +macro_rules! setup_db { + ( + $db_path:expr, + $a_timeout:expr, + $opts:expr, + $m_name:literal $(,)? + ) => {{ + async fn migrate(pool: &SqlitePool) -> sqlx::Result<()> { + { sqlx::sqlx_macros::migrate!($m_name) }.run(pool).await?; + Ok(()) + } + + crate::aclient::utils::setup_db_inner($db_path, $a_timeout, $opts, migrate) + }}; +} +pub(crate) use setup_db; + +use std::{os::fd::AsRawFd, path::Path, time::Duration}; + +use fs_err::OpenOptions; +use sqlx::{ + SqlitePool, + sqlite::{SqliteConnectOptions, SqlitePoolOptions}, +}; +use tracing::debug; + +/// Helper for `setup_db!` +pub(crate) async fn setup_db_inner( + db_path: &Path, + acquire_timeout: f64, + mk_opts: fn(&str) -> sqlx::Result<SqliteConnectOptions>, + migrate: impl AsyncFn(&SqlitePool) -> sqlx::Result<()>, +) -> sqlx::Result<SqlitePool> { + async fn open_db(timeout: f64, opts: SqliteConnectOptions) -> sqlx::Result<SqlitePool> { + let pool = SqlitePoolOptions::new() + .acquire_timeout(Duration::from_secs_f64(timeout)) + .connect_with(opts) + .await?; + + Ok(pool) + } + + { + let file = OpenOptions::new() + .read(true) + .write(true) + .create(true) + .open(db_path)?; + + // Lock the db file while we are running the migrations. + // Why? Because there is a small chance that we start running migrations (e.g. as the daemon) + // and then another process is started, which will also try to run migrations. + // Essentially, one of the processes will receive with a SQLite UNIQUE constraint failure. + // So let's avoid that possibility from the start. + file.lock()?; + + let pool = open_db( + acquire_timeout, + mk_opts(format!("/proc/self/fd/{}", file.as_raw_fd()).as_str())?, + ) + .await?; + + debug!("running sqlite database setup"); + + migrate(&pool).await?; + + file.unlock()?; + } + + let real_opts = mk_opts(db_path.to_str().expect("Should be utf-8"))?; + let pool = open_db(acquire_timeout, real_opts).await?; + + Ok(pool) +} |
