chore: Commit

3 files changed, 601 insertions, 0 deletions

diff --git a/crates/turtle/src/history/builder.rs b/crates/turtle/src/history/builder.rs
new file mode 100644
index 00000000..7eca0491
--- /dev/null
+++ b/crates/turtle/src/history/builder.rs
@@ -0,0 +1,78 @@
+use typed_builder::TypedBuilder;
+
+use super::History;
+
+/// Builder for a history entry that is loaded from the database.
+///
+/// All fields are required, as they are all present in the database.
+#[derive(Debug, Clone, TypedBuilder)]
+pub struct HistoryFromDb {
+    id: String,
+    timestamp: time::OffsetDateTime,
+    command: String,
+    cwd: String,
+    exit: i64,
+    duration: i64,
+    session: String,
+    hostname: String,
+    author: String,
+    intent: Option<String>,
+    deleted_at: Option<time::OffsetDateTime>,
+}
+
+impl From<HistoryFromDb> for History {
+    fn from(from_db: HistoryFromDb) -> Self {
+        Self {
+            id: from_db.id.into(),
+            timestamp: from_db.timestamp,
+            exit: from_db.exit,
+            command: from_db.command,
+            cwd: from_db.cwd,
+            duration: from_db.duration,
+            session: from_db.session,
+            hostname: from_db.hostname,
+            author: from_db.author,
+            intent: from_db.intent,
+            deleted_at: from_db.deleted_at,
+        }
+    }
+}
+
+/// Builder for a history entry that is captured via hook and sent to the daemon
+///
+/// This builder is similar to Capture, but we just require more information up front.
+/// For the old setup, we could just rely on `History::new` to read some of the missing
+/// data. This is no longer the case.
+#[derive(Debug, Clone, TypedBuilder)]
+pub struct HistoryDaemonCapture {
+    timestamp: time::OffsetDateTime,
+    #[builder(setter(into))]
+    command: String,
+    #[builder(setter(into))]
+    cwd: String,
+    #[builder(setter(into))]
+    session: String,
+    #[builder(setter(into))]
+    hostname: String,
+    #[builder(default, setter(strip_option, into))]
+    author: Option<String>,
+    #[builder(default, setter(strip_option, into))]
+    intent: Option<String>,
+}
+
+impl From<HistoryDaemonCapture> for History {
+    fn from(captured: HistoryDaemonCapture) -> Self {
+        Self::new(
+            captured.timestamp,
+            captured.command,
+            captured.cwd,
+            -1,
+            -1,
+            Some(captured.session),
+            Some(captured.hostname),
+            captured.author,
+            captured.intent,
+            None,
+        )
+    }
+}
diff --git a/crates/turtle/src/history/mod.rs b/crates/turtle/src/history/mod.rs
new file mode 100644
index 00000000..10e74d8e
--- /dev/null
+++ b/crates/turtle/src/history/mod.rs
@@ -0,0 +1,300 @@
+use core::fmt::Formatter;
+use regex::RegexSet;
+use std::env;
+use std::fmt::Display;
+
+use turtle_common::utils::uuid_v7;
+
+use time::OffsetDateTime;
+
+use crate::history::secrets::SECRET_PATTERNS_RE;
+
+pub mod builder;
+mod secrets;
+
+const HISTORY_AUTHOR_ENV: &str = "ATUIN_HISTORY_AUTHOR";
+const HISTORY_INTENT_ENV: &str = "ATUIN_HISTORY_INTENT";
+
+#[derive(Clone, Debug, Eq, PartialEq, Hash)]
+pub struct HistoryId(pub String);
+
+impl Display for HistoryId {
+    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+        write!(f, "{}", self.0)
+    }
+}
+
+impl From<String> for HistoryId {
+    fn from(s: String) -> Self {
+        Self(s)
+    }
+}
+
+pub(crate) fn get_hostname() -> String {
+    env::var("ATUIN_HOST_NAME")
+        .unwrap_or_else(|_| whoami::hostname().unwrap_or_else(|_| "unknown-host".to_string()))
+}
+
+pub(crate) fn get_username() -> String {
+    env::var("ATUIN_HOST_USER")
+        .unwrap_or_else(|_| whoami::username().unwrap_or_else(|_| "unknown-user".to_string()))
+}
+
+/// Returns a pair of the hostname and username, separated by a colon.
+#[must_use]
+pub fn get_host_user() -> String {
+    format!("{}:{}", get_hostname(), get_username())
+}
+
+/// Client-side history entry.
+///
+/// Client stores data unencrypted, and only encrypts it before sending to the server.
+///
+/// To create a new history entry, use one of the builders:
+/// - [`History::import()`] to import an entry from the shell history file
+/// - [`History::capture()`] to capture an entry via hook
+/// - [`History::from_db()`] to create an instance from the database entry
+//
+// ## Implementation Notes
+//
+// New fields must be added to `History::{serialize,deserialize}` in a backwards
+// compatible way (sensible defaults and careful `nfields` handling).
+#[derive(Debug, Clone, PartialEq, Eq, sqlx::FromRow)]
+pub struct History {
+    /// A client-generated ID, used to identify the entry when syncing.
+    ///
+    /// Stored as `client_id` in the database.
+    pub id: HistoryId,
+
+    /// When the command was run.
+    pub timestamp: OffsetDateTime,
+
+    /// How long the command took to run.
+    pub duration: i64,
+
+    /// The exit code of the command.
+    pub exit: i64,
+
+    /// The command that was run.
+    pub command: String,
+
+    /// The current working directory when the command was run.
+    pub cwd: String,
+
+    /// The session ID, associated with a terminal session.
+    pub session: String,
+
+    /// The hostname of the machine the command was run on.
+    pub hostname: String,
+
+    /// Who wrote this command (human user or automation/agent identity).
+    pub author: String,
+
+    /// Optional rationale for why the command was executed.
+    pub intent: Option<String>,
+
+    /// Timestamp, which is set when the entry is deleted, allowing a soft delete.
+    pub deleted_at: Option<OffsetDateTime>,
+}
+
+impl History {
+    #[must_use]
+    pub fn author_from_hostname(hostname: &str) -> String {
+        hostname
+            .split_once(':')
+            .map_or_else(|| hostname.to_owned(), |(_, user)| user.to_owned())
+    }
+
+    fn normalize_optional_field(field: Option<String>) -> Option<String> {
+        field.and_then(|value| {
+            let trimmed = value.trim();
+            if trimmed.is_empty() {
+                None
+            } else {
+                Some(trimmed.to_owned())
+            }
+        })
+    }
+
+    #[expect(clippy::too_many_arguments)]
+    fn new(
+        timestamp: OffsetDateTime,
+        command: String,
+        cwd: String,
+        exit: i64,
+        duration: i64,
+        session: Option<String>,
+        hostname: Option<String>,
+        author: Option<String>,
+        intent: Option<String>,
+        deleted_at: Option<OffsetDateTime>,
+    ) -> Self {
+        let session = session
+            .or_else(|| env::var("ATUIN_SESSION").ok())
+            .unwrap_or_else(|| uuid_v7().as_simple().to_string());
+        let hostname = hostname.unwrap_or_else(get_host_user);
+        let author = Self::normalize_optional_field(author)
+            .or_else(|| Self::normalize_optional_field(env::var(HISTORY_AUTHOR_ENV).ok()))
+            .unwrap_or_else(|| Self::author_from_hostname(hostname.as_str()));
+        let intent = Self::normalize_optional_field(intent)
+            .or_else(|| Self::normalize_optional_field(env::var(HISTORY_INTENT_ENV).ok()));
+
+        Self {
+            id: uuid_v7().as_simple().to_string().into(),
+            timestamp,
+            command,
+            cwd,
+            exit,
+            duration,
+            session,
+            hostname,
+            author,
+            intent,
+            deleted_at,
+        }
+    }
+
+    /// Builder for a history entry that is captured via hook, and sent to the daemon.
+    ///
+    /// This builder is used only at the `start` step of the hook,
+    /// so it doesn't have any fields which are known only after
+    /// the command is finished, such as `exit` or `duration`.
+    ///
+    /// It does, however, include information that can usually be inferred.
+    ///
+    /// This is because the daemon we are sending a request to lacks the context of the command
+    ///
+    /// ## Examples
+    /// ```rust
+    /// use crate::aclient::history::History;
+    ///
+    /// let history: History = History::daemon()
+    ///     .timestamp(time::OffsetDateTime::now_utc())
+    ///     .command("ls -la")
+    ///     .cwd("/home/user")
+    ///     .session("018deb6e8287781f9973ef40e0fde76b")
+    ///     .hostname("computer:ellie")
+    ///     .build()
+    ///     .into();
+    /// ```
+    ///
+    /// Command without any required info cannot be captured, which is forced at compile time:
+    ///
+    /// ```compile_fail
+    /// use crate::aclient::history::History;
+    ///
+    /// // this will not compile because `hostname` is missing
+    /// let history: History = History::daemon()
+    ///     .timestamp(time::OffsetDateTime::now_utc())
+    ///     .command("ls -la")
+    ///     .cwd("/home/user")
+    ///     .session("018deb6e8287781f9973ef40e0fde76b")
+    ///     .build()
+    ///     .into();
+    /// ```
+    pub fn daemon() -> builder::HistoryDaemonCaptureBuilder {
+        builder::HistoryDaemonCapture::builder()
+    }
+
+    #[doc(hidden)]
+    pub fn from_db() -> builder::HistoryFromDbBuilder {
+        builder::HistoryFromDb::builder()
+    }
+
+    pub fn should_save(&self, filter: SettingsFilter<'_>) -> bool {
+        !(self.command.is_empty()
+            || filter.history.is_match(&self.command)
+            || filter.cwd.is_match(&self.cwd)
+            || (filter.secrets && SECRET_PATTERNS_RE.is_match(&self.command)))
+    }
+}
+
+#[derive(Debug, Copy, Clone)]
+pub struct SettingsFilter<'a> {
+    pub history: &'a RegexSet,
+    pub cwd: &'a RegexSet,
+    pub secrets: bool,
+}
+
+#[cfg(test)]
+mod tests {
+    // use regex::RegexSet;
+    //
+    // use crate::history::History;
+
+    // // Test that we don't save history where necessary
+    // #[test]
+    // fn privacy_test() {
+    //     let settings = Settings {
+    //         cwd_filter: RegexSet::new(["^/supasecret"]).unwrap(),
+    //         history_filter: RegexSet::new(["^psql"]).unwrap(),
+    //         ..Settings::default()
+    //     };
+    //
+    //     let normal_command: History = History::daemon()
+    //         .timestamp(time::OffsetDateTime::now_utc())
+    //         .command("echo foo")
+    //         .cwd("/")
+    //         .build()
+    //         .into();
+    //
+    //     let with_space: History = History::daemon()
+    //         .timestamp(time::OffsetDateTime::now_utc())
+    //         .command(" echo bar")
+    //         .cwd("/")
+    //         .build()
+    //         .into();
+    //
+    //     let empty: History = History::daemon()
+    //         .timestamp(time::OffsetDateTime::now_utc())
+    //         .command("")
+    //         .cwd("/")
+    //         .build()
+    //         .into();
+    //
+    //     let stripe_key: History = History::daemon()
+    //         .timestamp(time::OffsetDateTime::now_utc())
+    //         .command("curl foo.com/bar?key=sk_test_1234567890abcdefghijklmnop")
+    //         .cwd("/")
+    //         .build()
+    //         .into();
+    //
+    //     let secret_dir: History = History::daemon()
+    //         .timestamp(time::OffsetDateTime::now_utc())
+    //         .command("echo ohno")
+    //         .cwd("/supasecret")
+    //         .build()
+    //         .into();
+    //
+    //     let with_psql: History = History::daemon()
+    //         .timestamp(time::OffsetDateTime::now_utc())
+    //         .command("psql")
+    //         .cwd("/supasecret")
+    //         .build()
+    //         .into();
+    //
+    //     assert!(normal_command.should_save(&settings));
+    //     assert!(!with_space.should_save(&settings));
+    //     assert!(!empty.should_save(&settings));
+    //     assert!(!stripe_key.should_save(&settings));
+    //     assert!(!secret_dir.should_save(&settings));
+    //     assert!(!with_psql.should_save(&settings));
+    // }
+    //
+    // #[test]
+    // fn disable_secrets() {
+    //     let settings = Settings {
+    //         secrets_filter: false,
+    //         ..Settings::new().unwrap()
+    //     };
+    //
+    //     let stripe_key: History = History::capture()
+    //         .timestamp(time::OffsetDateTime::now_utc())
+    //         .command("curl foo.com/bar?key=sk_test_1234567890abcdefghijklmnop")
+    //         .cwd("/")
+    //         .build()
+    //         .into();
+    //
+    //     assert!(stripe_key.should_save(&settings));
+    // }
+}
diff --git a/crates/turtle/src/history/secrets.rs b/crates/turtle/src/history/secrets.rs
new file mode 100644
index 00000000..08d24339
--- /dev/null
+++ b/crates/turtle/src/history/secrets.rs
@@ -0,0 +1,223 @@
+// This file will probably trigger a lot of scanners. Sorry.
+
+use regex::RegexSet;
+use std::sync::LazyLock;
+
+#[cfg(test)]
+pub(crate) enum TestValue<'a> {
+    Single(&'a str),
+    Multiple(&'a [&'a str]),
+}
+
+#[cfg(test)]
+type SpType<'a> = &'a [(&'a str, &'a str, TestValue<'a>)];
+
+#[cfg(not(test))]
+type SpType<'a> = &'a [(&'a str, &'a str)];
+
+/// A list of `(name, regex, test)`, where `test` should match against `regex`.
+pub(crate) static SECRET_PATTERNS: SpType<'_> = &[
+    (
+        "AWS Access Key ID",
+        "A[KS]IA[0-9A-Z]{16}",
+        #[cfg(test)]
+        TestValue::Single("AKIAIOSFODNN7EXAMPLE"),
+    ),
+    (
+        "AWS Secret Access Key env var",
+        "AWS_SECRET_ACCESS_KEY",
+        #[cfg(test)]
+        TestValue::Single("AWS_SECRET_ACCESS_KEY=KEYDATA"),
+    ),
+    (
+        "AWS Session Token env var",
+        "AWS_SESSION_TOKEN",
+        #[cfg(test)]
+        TestValue::Single("AWS_SESSION_TOKEN=KEYDATA"),
+    ),
+    (
+        "Microsoft Azure secret access key env var",
+        "AZURE_.*_KEY",
+        #[cfg(test)]
+        TestValue::Single("export AZURE_STORAGE_ACCOUNT_KEY=KEYDATA"),
+    ),
+    (
+        "Google cloud platform key env var",
+        "GOOGLE_SERVICE_ACCOUNT_KEY",
+        #[cfg(test)]
+        TestValue::Single("export GOOGLE_SERVICE_ACCOUNT_KEY=KEYDATA"),
+    ),
+    (
+        "Atuin login",
+        r"atuin\s+login",
+        #[cfg(test)]
+        TestValue::Single(
+            "atuin login -u mycoolusername -p mycoolpassword -k \"lots of random words\"",
+        ),
+    ),
+    (
+        "GitHub PAT (old)",
+        "ghp_[a-zA-Z0-9]{36}",
+        #[cfg(test)]
+        TestValue::Single("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"), // legit, I expired it
+    ),
+    (
+        "GitHub PAT (new)",
+        "gh1_[A-Za-z0-9]{21}_[A-Za-z0-9]{59}|github_pat_[0-9][A-Za-z0-9]{21}_[A-Za-z0-9]{59}",
+        #[cfg(test)]
+        TestValue::Multiple(&[
+            "gh1_1234567890abcdefghijk_1234567890abcdefghijklmnopqrstuvwxyz1234567890abcdefghijklm",
+            "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", // also legit, also expired
+        ]),
+    ),
+    (
+        "GitHub OAuth Access Token",
+        "gho_[A-Za-z0-9]{36}",
+        #[cfg(test)]
+        TestValue::Single("gho_1234567890abcdefghijklmnopqrstuvwx000"), // not a real token
+    ),
+    (
+        "GitHub OAuth Access Token (user)",
+        "ghu_[A-Za-z0-9]{36}",
+        #[cfg(test)]
+        TestValue::Single("ghu_1234567890abcdefghijklmnopqrstuvwx000"), // not a real token
+    ),
+    (
+        "GitHub App Installation Access Token",
+        "ghs_[A-Za-z0-9._-]{36,}",
+        #[cfg(test)]
+        TestValue::Multiple(&[
+            "ghs_1234567890abcdefghijklmnopqrstuvwx000", // not a real token
+            "ghs_abc-def.ghi_jklMNOP0123456789qrstuv-wxyzABCD", // new token format, fake data
+        ]),
+    ),
+    (
+        "GitHub Refresh Token",
+        "ghr_[A-Za-z0-9]{76}",
+        #[cfg(test)]
+        TestValue::Single(
+            "ghr_1234567890abcdefghijklmnopqrstuvwx1234567890abcdefghijklmnopqrstuvwx1234567890abcdefghijklmnopqrstuvwx",
+        ), // not a real token
+    ),
+    (
+        "GitHub App Installation Access Token v1",
+        "v1\\.[0-9A-Fa-f]{40}",
+        #[cfg(test)]
+        TestValue::Single("v1.1234567890abcdef1234567890abcdef12345678"), // not a real token
+    ),
+    (
+        "GitLab PAT",
+        "glpat-[a-zA-Z0-9_]{20}",
+        #[cfg(test)]
+        TestValue::Single("glpat-RkE_BG5p_bbjML21WSfy"),
+    ),
+    (
+        "Slack OAuth v2 bot",
+        "xoxb-[0-9]{11}-[0-9]{11}-[0-9a-zA-Z]{24}",
+        #[cfg(test)]
+        TestValue::Single("xoxb-17653672481-19874698323-pdFZKVeTuE8sk7oOcBrzbqgy"),
+    ),
+    (
+        "Slack OAuth v2 user token",
+        "xoxp-[0-9]{11}-[0-9]{11}-[0-9a-zA-Z]{24}",
+        #[cfg(test)]
+        TestValue::Single("xoxp-17653672481-19874698323-pdFZKVeTuE8sk7oOcBrzbqgy"),
+    ),
+    (
+        "Slack webhook",
+        "T[a-zA-Z0-9_]{8}/B[a-zA-Z0-9_]{8}/[a-zA-Z0-9_]{24}",
+        #[cfg(test)]
+        TestValue::Single(
+            "https://hooks.slack.com/services/T00000000/B00000000/XXXXXXXXXXXXXXXXXXXXXXXX",
+        ),
+    ),
+    (
+        "Stripe test key",
+        "sk_test_[0-9a-zA-Z]{24}",
+        #[cfg(test)]
+        TestValue::Single("sk_test_1234567890abcdefghijklmnop"),
+    ),
+    (
+        "Stripe live key",
+        "sk_live_[0-9a-zA-Z]{24}",
+        #[cfg(test)]
+        TestValue::Single("sk_live_1234567890abcdefghijklmnop"),
+    ),
+    (
+        "Netlify authentication token",
+        "nf[pcoub]_[0-9a-zA-Z]{36}",
+        #[cfg(test)]
+        TestValue::Single("nfp_nBh7BdJxUwyaBBwFzpyD29MMFT6pZ9wq5634"),
+    ),
+    (
+        "npm token",
+        "npm_[A-Za-z0-9]{36}",
+        #[cfg(test)]
+        TestValue::Single("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"),
+    ),
+    (
+        "Pulumi personal access token",
+        "pul-[0-9a-f]{40}",
+        #[cfg(test)]
+        TestValue::Single("pul-683c2770662c51d960d72ec27613be7653c5cb26"),
+    ),
+];
+
+/// The `regex` expressions from [`SECRET_PATTERNS`] compiled into a `RegexSet`.
+pub(crate) static SECRET_PATTERNS_RE: LazyLock<RegexSet> = LazyLock::new(|| {
+    let exprs = SECRET_PATTERNS.iter().map(|f| f.1);
+    RegexSet::new(exprs).expect("Failed to build secrets regex")
+});
+
+#[cfg(test)]
+mod tests {
+    use regex::Regex;
+
+    use crate::aclient::secrets::{SECRET_PATTERNS, TestValue};
+
+    #[test]
+    fn test_secrets() {
+        for (name, regex, test) in SECRET_PATTERNS {
+            let re =
+                Regex::new(regex).unwrap_or_else(|_| panic!("Failed to compile regex for {name}"));
+
+            match test {
+                TestValue::Single(test) => {
+                    assert!(re.is_match(test), "{name} test failed!");
+                }
+                TestValue::Multiple(tests) => {
+                    for test_str in tests.iter() {
+                        assert!(
+                            re.is_match(test_str),
+                            "{name} test with value \"{test_str}\" failed!"
+                        );
+                    }
+                }
+            }
+        }
+    }
+
+    #[test]
+    fn test_secrets_embedded() {
+        for (name, regex, test) in SECRET_PATTERNS {
+            let re =
+                Regex::new(regex).unwrap_or_else(|_| panic!("Failed to compile regex for {name}"));
+
+            match test {
+                TestValue::Single(test) => {
+                    let embedded = format!("some random text {test} some more random text");
+                    assert!(re.is_match(&embedded), "{name} embedded test failed!");
+                }
+                TestValue::Multiple(tests) => {
+                    for test_str in tests.iter() {
+                        let embedded = format!("some random text {test_str} some more random text");
+                        assert!(
+                            re.is_match(&embedded),
+                            "{name} embedded test with value \"{test_str}\" failed!"
+                        );
+                    }
+                }
+            }
+        }
+    }
+}