Importer V3 (#395)

* start of importer refactor

* fish

* resh

* zsh

Cargo.lock

@@ -118,6 +118,7 @@ dependencies = [
  "itertools",
  "lazy_static",
  "log",
+ "memchr",
  "minspan",
  "parse_duration",
  "regex",
@@ -1189,9 +1190,9 @@ dependencies = [
 
 [[package]]
 name = "memchr"
-version = "2.4.1"
+version = "2.5.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "308cc39be01b73d0d18f82a0e7b2a3df85245f84af96fdddc5d202d27e47b86a"
+checksum = "2dffe52ecf27772e601905b7522cb4ef790d2cc203488bbd0e2fe85fcb74566d"
 
 [[package]]
 name = "mime"

atuin-client/Cargo.toml

@@ -49,6 +49,7 @@ regex = "1.5.4"
 fs-err = "2.7"
 sql-builder = "3"
 lazy_static = "1"
+memchr = "2.5"
 
 # sync
 urlencoding = { version = "2.1.0", optional = true }

atuin-client/src/database.rs

@@ -41,7 +41,7 @@ pub fn current_context() -> Context {
 }
 
 #[async_trait]
-pub trait Database {
+pub trait Database: Send + Sync {
     async fn save(&mut self, h: &History) -> Result<()>;
     async fn save_bulk(&mut self, h: &[History]) -> Result<()>;
 

atuin-client/src/import/bash.rs

@@ -1,134 +1,106 @@
-use std::{
+use std::{fs::File, io::Read, path::PathBuf};
-    fs::File,
-    io::{BufRead, BufReader, Read, Seek},
-    path::{Path, PathBuf},
-};
 
+use async_trait::async_trait;
 use directories::UserDirs;
 use eyre::{eyre, Result};
 
-use super::{count_lines, Importer};
+use super::{get_histpath, unix_byte_lines, Importer, Loader};
 use crate::history::History;
 
 #[derive(Debug)]
-pub struct Bash<R> {
+pub struct Bash {
-    file: BufReader<R>,
+    bytes: Vec<u8>,
-    strbuf: String,
-    loc: usize,
-    counter: i64,
 }
 
-impl<R: Read + Seek> Bash<R> {
+fn default_histpath() -> Result<PathBuf> {
-    fn new(r: R) -> Result<Self> {
+    let user_dirs = UserDirs::new().ok_or_else(|| eyre!("could not find user directories"))?;
-        let mut buf = BufReader::new(r);
+    let home_dir = user_dirs.home_dir();
-        let loc = count_lines(&mut buf)?;
 
-        Ok(Self {
+    Ok(home_dir.join(".bash_history"))
-            file: buf,
-            strbuf: String::new(),
-            loc,
-            counter: 0,
-        })
-    }
 }
 
-impl Importer for Bash<File> {
+#[async_trait]
+impl Importer for Bash {
     const NAME: &'static str = "bash";
 
-    fn histpath() -> Result<PathBuf> {
+    async fn new() -> Result<Self> {
-        let user_dirs = UserDirs::new().unwrap();
+        let mut bytes = Vec::new();
-        let home_dir = user_dirs.home_dir();
+        let path = get_histpath(default_histpath)?;
-
+        let mut f = File::open(path)?;
-        Ok(home_dir.join(".bash_history"))
+        f.read_to_end(&mut bytes)?;
+        Ok(Self { bytes })
     }
 
-    fn parse(path: impl AsRef<Path>) -> Result<Self> {
+    async fn entries(&mut self) -> Result<usize> {
-        Self::new(File::open(path)?)
+        Ok(super::count_lines(&self.bytes))
     }
-}
 
-impl<R: Read> Iterator for Bash<R> {
+    async fn load(self, h: &mut impl Loader) -> Result<()> {
-    type Item = Result<History>;
+        let now = chrono::Utc::now();
+        let mut line = String::new();
 
-    fn next(&mut self) -> Option<Self::Item> {
+        for (i, b) in unix_byte_lines(&self.bytes).enumerate() {
-        self.strbuf.clear();
+            let s = match std::str::from_utf8(b) {
-        match self.file.read_line(&mut self.strbuf) {
+                Ok(s) => s,
-            Ok(0) => return None,
+                Err(_) => continue, // we can skip past things like invalid utf8
-            Ok(_) => (),
-            Err(e) => return Some(Err(eyre!("failed to read line: {}", e))), // we can skip past things like invalid utf8
-        }
-
-        self.loc -= 1;
-
-        while self.strbuf.ends_with("\\\n") {
-            if self.file.read_line(&mut self.strbuf).is_err() {
-                // There's a chance that the last line of a command has invalid
-                // characters, the only safe thing to do is break :/
-                // usually just invalid utf8 or smth
-                // however, we really need to avoid missing history, so it's
-                // better to have some items that should have been part of
-                // something else, than to miss things. So break.
-                break;
             };
 
-            self.loc -= 1;
+            if let Some(s) = s.strip_suffix('\\') {
+                line.push_str(s);
+                line.push_str("\\\n");
+            } else {
+                line.push_str(s);
+                let command = std::mem::take(&mut line);
+
+                let offset = chrono::Duration::seconds(i as i64);
+                h.push(History::new(
+                    now - offset, // preserve ordering
+                    command,
+                    String::from("unknown"),
+                    -1,
+                    -1,
+                    None,
+                    None,
+                ))
+                .await?;
+            }
         }
 
-        let time = chrono::Utc::now();
+        Ok(())
-        let offset = chrono::Duration::seconds(self.counter);
-        let time = time - offset;
-
-        self.counter += 1;
-
-        Some(Ok(History::new(
-            time,
-            self.strbuf.trim_end().to_string(),
-            String::from("unknown"),
-            -1,
-            -1,
-            None,
-            None,
-        )))
-    }
-
-    fn size_hint(&self) -> (usize, Option<usize>) {
-        (0, Some(self.loc))
     }
 }
 
 #[cfg(test)]
 mod tests {
-    use std::io::Cursor;
+    use itertools::assert_equal;
+
+    use crate::import::{tests::TestLoader, Importer};
 
     use super::Bash;
 
-    #[test]
+    #[tokio::test]
-    fn test_parse_file() {
+    async fn test_parse_file() {
-        let input = r"cargo install atuin
+        let bytes = r"cargo install atuin
 cargo install atuin; \
 cargo update
 cargo :b̷i̶t̴r̵o̴t̴ ̵i̷s̴ ̷r̶e̵a̸l̷
-";
+"
+        .as_bytes()
+        .to_owned();
 
-        let cursor = Cursor::new(input);
+        let mut bash = Bash { bytes };
-        let mut bash = Bash::new(cursor).unwrap();
+        assert_eq!(bash.entries().await.unwrap(), 4);
-        assert_eq!(bash.loc, 4);
-        assert_eq!(bash.size_hint(), (0, Some(4)));
 
-        assert_eq!(
+        let mut loader = TestLoader::default();
-            &bash.next().unwrap().unwrap().command,
+        bash.load(&mut loader).await.unwrap();
-            "cargo install atuin"
-        );
-        assert_eq!(
-            &bash.next().unwrap().unwrap().command,
-            "cargo install atuin; \\\ncargo update"
-        );
-        assert_eq!(
-            &bash.next().unwrap().unwrap().command,
-            "cargo :b̷i̶t̴r̵o̴t̴ ̵i̷s̴ ̷r̶e̵a̸l̷"
-        );
-        assert!(bash.next().is_none());
 
-        assert_eq!(bash.size_hint(), (0, Some(0)));
+        assert_equal(
+            loader.buf.iter().map(|h| h.command.as_str()),
+            [
+                "cargo install atuin",
+                "cargo install atuin; \\\ncargo update",
+                "cargo :b̷i̶t̴r̵o̴t̴ ̵i̷s̴ ̷r̶e̵a̸l̷",
+            ],
+        );
     }
 }

atuin-client/src/import/fish.rs

@@ -1,99 +1,90 @@
 // import old shell history!
 // automatically hoover up all that we can find
 
-use std::{
+use std::{fs::File, io::Read, path::PathBuf};
-    fs::File,
-    io::{self, BufRead, BufReader, Read, Seek},
-    path::{Path, PathBuf},
-};
 
+use async_trait::async_trait;
 use chrono::{prelude::*, Utc};
 use directories::BaseDirs;
 use eyre::{eyre, Result};
 
-use super::{count_lines, Importer};
+use super::{get_histpath, unix_byte_lines, Importer, Loader};
 use crate::history::History;
 
 #[derive(Debug)]
-pub struct Fish<R> {
+pub struct Fish {
-    file: BufReader<R>,
+    bytes: Vec<u8>,
-    strbuf: String,
-    loc: usize,
 }
 
-impl<R: Read + Seek> Fish<R> {
+/// see https://fishshell.com/docs/current/interactive.html#searchable-command-history
-    fn new(r: R) -> Result<Self> {
+fn default_histpath() -> Result<PathBuf> {
-        let mut buf = BufReader::new(r);
+    let base = BaseDirs::new().ok_or_else(|| eyre!("could not determine data directory"))?;
-        let loc = count_lines(&mut buf)?;
+    let data = base.data_local_dir();
 
-        Ok(Self {
+    // fish supports multiple history sessions
-            file: buf,
+    // If `fish_history` var is missing, or set to `default`, use `fish` as the session
-            strbuf: String::new(),
+    let session = std::env::var("fish_history").unwrap_or_else(|_| String::from("fish"));
-            loc,
+    let session = if session == "default" {
-        })
+        String::from("fish")
+    } else {
+        session
+    };
+
+    let mut histpath = data.join("fish");
+    histpath.push(format!("{}_history", session));
+
+    if histpath.exists() {
+        Ok(histpath)
+    } else {
+        Err(eyre!("Could not find history file. Try setting $HISTFILE"))
     }
 }
 
-impl<R: Read> Fish<R> {
+#[async_trait]
-    fn new_entry(&mut self) -> io::Result<bool> {
+impl Importer for Fish {
-        let inner = self.file.fill_buf()?;
-        Ok(inner.starts_with(b"- "))
-    }
-}
-
-impl Importer for Fish<File> {
     const NAME: &'static str = "fish";
 
-    /// see https://fishshell.com/docs/current/interactive.html#searchable-command-history
+    async fn new() -> Result<Self> {
-    fn histpath() -> Result<PathBuf> {
+        let mut bytes = Vec::new();
-        let base = BaseDirs::new().ok_or_else(|| eyre!("could not determine data directory"))?;
+        let path = get_histpath(default_histpath)?;
-        let data = base.data_local_dir();
+        let mut f = File::open(path)?;
-
+        f.read_to_end(&mut bytes)?;
-        // fish supports multiple history sessions
+        Ok(Self { bytes })
-        // If `fish_history` var is missing, or set to `default`, use `fish` as the session
-        let session = std::env::var("fish_history").unwrap_or_else(|_| String::from("fish"));
-        let session = if session == "default" {
-            String::from("fish")
-        } else {
-            session
-        };
-
-        let mut histpath = data.join("fish");
-        histpath.push(format!("{}_history", session));
-
-        if histpath.exists() {
-            Ok(histpath)
-        } else {
-            Err(eyre!("Could not find history file. Try setting $HISTFILE"))
-        }
     }
 
-    fn parse(path: impl AsRef<Path>) -> Result<Self> {
+    async fn entries(&mut self) -> Result<usize> {
-        Self::new(File::open(path)?)
+        Ok(super::count_lines(&self.bytes))
     }
-}
 
-impl<R: Read> Iterator for Fish<R> {
+    async fn load(self, loader: &mut impl Loader) -> Result<()> {
-    type Item = Result<History>;
+        let now = Utc::now();
-
-    fn next(&mut self) -> Option<Self::Item> {
         let mut time: Option<DateTime<Utc>> = None;
         let mut cmd: Option<String> = None;
 
-        loop {
+        for b in unix_byte_lines(&self.bytes) {
-            self.strbuf.clear();
+            let s = match std::str::from_utf8(b) {
-            match self.file.read_line(&mut self.strbuf) {
+                Ok(s) => s,
-                // no more content to read
+                Err(_) => continue, // we can skip past things like invalid utf8
-                Ok(0) => break,
+            };
-                // bail on IO error
-                Err(e) => return Some(Err(e.into())),
-                _ => (),
-            }
 
-            // `read_line` adds the line delimeter to the string. No thanks
+            if let Some(c) = s.strip_prefix("- cmd: ") {
-            self.strbuf.pop();
+                // first, we must deal with the prev cmd
+                if let Some(cmd) = cmd.take() {
+                    let time = time.unwrap_or(now);
+
+                    loader
+                        .push(History::new(
+                            time,
+                            cmd,
+                            "unknown".into(),
+                            -1,
+                            -1,
+                            None,
+                            None,
+                        ))
+                        .await?;
+                }
 
-            if let Some(c) = self.strbuf.strip_prefix("- cmd: ") {
                 // using raw strings to avoid needing escaping.
                 // replaces double backslashes with single backslashes
                 let c = c.replace(r"\\", r"\");
@@ -102,7 +93,7 @@ impl<R: Read> Iterator for Fish<R> {
                 // TODO: any other escape characters?
 
                 cmd = Some(c);
-            } else if let Some(t) = self.strbuf.strip_prefix("  when: ") {
+            } else if let Some(t) = s.strip_prefix("  when: ") {
                 // if t is not an int, just ignore this line
                 if let Ok(t) = t.parse::<i64>() {
                     time = Some(Utc.timestamp(t, 0));
@@ -110,47 +101,40 @@ impl<R: Read> Iterator for Fish<R> {
             } else {
                 // ... ignore paths lines
             }
-
-            match self.new_entry() {
-                // next line is a new entry, so let's stop here
-                // only if we have found a command though
-                Ok(true) if cmd.is_some() => break,
-                // bail on IO error
-                Err(e) => return Some(Err(e.into())),
-                _ => (),
-            }
         }
 
-        let cmd = cmd?;
+        // we might have a trailing cmd
-        let time = time.unwrap_or_else(Utc::now);
+        if let Some(cmd) = cmd.take() {
+            let time = time.unwrap_or(now);
 
-        Some(Ok(History::new(
+            loader
-            time,
+                .push(History::new(
-            cmd,
+                    time,
-            "unknown".into(),
+                    cmd,
-            -1,
+                    "unknown".into(),
-            -1,
+                    -1,
-            None,
+                    -1,
-            None,
+                    None,
-        )))
+                    None,
-    }
+                ))
+                .await?;
+        }
 
-    fn size_hint(&self) -> (usize, Option<usize>) {
+        Ok(())
-        // worst case, entry per line
-        (0, Some(self.loc))
     }
 }
 
 #[cfg(test)]
 mod test {
-    use std::io::Cursor;
+
+    use crate::import::{tests::TestLoader, Importer};
 
     use super::Fish;
 
-    #[test]
+    #[tokio::test]
-    fn parse_complex() {
+    async fn parse_complex() {
         // complicated input with varying contents and escaped strings.
-        let input = r#"- cmd: history --help
+        let bytes = r#"- cmd: history --help
   when: 1639162832
 - cmd: cat ~/.bash_history
   when: 1639162851
@@ -181,14 +165,20 @@ ERROR
   when: 1639163066
   paths:
     - ~/.local/share/fish/fish_history
-"#;
+"#
-        let cursor = Cursor::new(input);
+        .as_bytes()
-        let mut fish = Fish::new(cursor).unwrap();
+        .to_owned();
+
+        let fish = Fish { bytes };
+
+        let mut loader = TestLoader::default();
+        fish.load(&mut loader).await.unwrap();
+        let mut history = loader.buf.into_iter();
 
         // simple wrapper for fish history entry
         macro_rules! fishtory {
             ($timestamp:expr, $command:expr) => {
-                let h = fish.next().expect("missing entry in history").unwrap();
+                let h = history.next().expect("missing entry in history");
                 assert_eq!(h.command.as_str(), $command);
                 assert_eq!(h.timestamp.timestamp(), $timestamp);
             };

atuin-client/src/import/mod.rs

@@ -1,9 +1,8 @@
-use std::{
+use std::path::PathBuf;
-    io::{BufRead, BufReader, Read, Seek, SeekFrom},
-    path::{Path, PathBuf},
-};
 
-use eyre::Result;
+use async_trait::async_trait;
+use eyre::{bail, Result};
+use memchr::Memchr;
 
 use crate::history::History;
 
@@ -12,16 +11,88 @@ pub mod fish;
 pub mod resh;
 pub mod zsh;
 
-// this could probably be sped up
+#[async_trait]
-fn count_lines(buf: &mut BufReader<impl Read + Seek>) -> Result<usize> {
+pub trait Importer: Sized {
-    let lines = buf.lines().count();
-    buf.seek(SeekFrom::Start(0))?;
-
-    Ok(lines)
-}
-
-pub trait Importer: IntoIterator<Item = Result<History>> + Sized {
     const NAME: &'static str;
-    fn histpath() -> Result<PathBuf>;
+    async fn new() -> Result<Self>;
-    fn parse(path: impl AsRef<Path>) -> Result<Self>;
+    async fn entries(&mut self) -> Result<usize>;
+    async fn load(self, loader: &mut impl Loader) -> Result<()>;
+}
+
+#[async_trait]
+pub trait Loader: Sync + Send {
+    async fn push(&mut self, hist: History) -> eyre::Result<()>;
+}
+
+fn unix_byte_lines(input: &[u8]) -> impl Iterator<Item = &[u8]> {
+    UnixByteLines {
+        iter: memchr::memchr_iter(b'\n', input),
+        bytes: input,
+        i: 0,
+    }
+}
+
+struct UnixByteLines<'a> {
+    iter: Memchr<'a>,
+    bytes: &'a [u8],
+    i: usize,
+}
+
+impl<'a> Iterator for UnixByteLines<'a> {
+    type Item = &'a [u8];
+
+    fn next(&mut self) -> Option<Self::Item> {
+        let j = self.iter.next()?;
+        let out = &self.bytes[self.i..j];
+        self.i = j + 1;
+        Some(out)
+    }
+
+    fn count(self) -> usize
+    where
+        Self: Sized,
+    {
+        self.iter.count()
+    }
+}
+
+fn count_lines(input: &[u8]) -> usize {
+    unix_byte_lines(input).count()
+}
+
+fn get_histpath<D>(def: D) -> Result<PathBuf>
+where
+    D: FnOnce() -> Result<PathBuf>,
+{
+    if let Ok(p) = std::env::var("HISTFILE") {
+        is_file(PathBuf::from(p))
+    } else {
+        is_file(def()?)
+    }
+}
+
+fn is_file(p: PathBuf) -> Result<PathBuf> {
+    if p.is_file() {
+        Ok(p)
+    } else {
+        bail!("Could not find history file {:?}. Try setting $HISTFILE", p)
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[derive(Default)]
+    pub struct TestLoader {
+        pub buf: Vec<History>,
+    }
+
+    #[async_trait]
+    impl Loader for TestLoader {
+        async fn push(&mut self, hist: History) -> Result<()> {
+            self.buf.push(hist);
+            Ok(())
+        }
+    }
 }

atuin-client/src/import/resh.rs

@@ -1,9 +1,6 @@
-use std::{
+use std::{fs::File, io::Read, path::PathBuf};
-    fs::File,
-    io::{BufRead, BufReader},
-    path::{Path, PathBuf},
-};
 
+use async_trait::async_trait;
 use chrono::{TimeZone, Utc};
 use directories::UserDirs;
 use eyre::{eyre, Result};
@@ -11,7 +8,7 @@ use serde::Deserialize;
 
 use atuin_common::utils::uuid_v4;
 
-use super::{count_lines, Importer};
+use super::{get_histpath, unix_byte_lines, Importer, Loader};
 use crate::history::History;
 
 #[derive(Deserialize, Debug)]
@@ -72,88 +69,72 @@ pub struct ReshEntry {
 
 #[derive(Debug)]
 pub struct Resh {
-    file: BufReader<File>,
+    bytes: Vec<u8>,
-    strbuf: String,
-    loc: usize,
 }
 
+fn default_histpath() -> Result<PathBuf> {
+    let user_dirs = UserDirs::new().ok_or_else(|| eyre!("could not find user directories"))?;
+    let home_dir = user_dirs.home_dir();
+
+    Ok(home_dir.join(".resh_history.json"))
+}
+
+#[async_trait]
 impl Importer for Resh {
     const NAME: &'static str = "resh";
 
-    fn histpath() -> Result<PathBuf> {
+    async fn new() -> Result<Self> {
-        let user_dirs = UserDirs::new().unwrap();
+        let mut bytes = Vec::new();
-        let home_dir = user_dirs.home_dir();
+        let path = get_histpath(default_histpath)?;
-
+        let mut f = File::open(path)?;
-        Ok(home_dir.join(".resh_history.json"))
+        f.read_to_end(&mut bytes)?;
+        Ok(Self { bytes })
     }
 
-    fn parse(path: impl AsRef<Path>) -> Result<Self> {
+    async fn entries(&mut self) -> Result<usize> {
-        let file = File::open(path)?;
+        Ok(super::count_lines(&self.bytes))
-        let mut buf = BufReader::new(file);
-        let loc = count_lines(&mut buf)?;
-
-        Ok(Self {
-            file: buf,
-            strbuf: String::new(),
-            loc,
-        })
     }
-}
 
-impl Iterator for Resh {
+    async fn load(self, h: &mut impl Loader) -> Result<()> {
-    type Item = Result<History>;
+        for b in unix_byte_lines(&self.bytes) {
+            let s = match std::str::from_utf8(b) {
+                Ok(s) => s,
+                Err(_) => continue, // we can skip past things like invalid utf8
+            };
+            let entry = match serde_json::from_str::<ReshEntry>(s) {
+                Ok(e) => e,
+                Err(_) => continue, // skip invalid json :shrug:
+            };
 
-    fn next(&mut self) -> Option<Self::Item> {
+            #[allow(clippy::cast_possible_truncation)]
-        self.strbuf.clear();
+            #[allow(clippy::cast_sign_loss)]
-        match self.file.read_line(&mut self.strbuf) {
+            let timestamp = {
-            Ok(0) => return None,
+                let secs = entry.realtime_before.floor() as i64;
-            Ok(_) => (),
+                let nanosecs = (entry.realtime_before.fract() * 1_000_000_000_f64).round() as u32;
-            Err(e) => return Some(Err(eyre!("failed to read line: {}", e))), // we can skip past things like invalid utf8
+                Utc.timestamp(secs, nanosecs)
+            };
+            #[allow(clippy::cast_possible_truncation)]
+            #[allow(clippy::cast_sign_loss)]
+            let duration = {
+                let secs = entry.realtime_after.floor() as i64;
+                let nanosecs = (entry.realtime_after.fract() * 1_000_000_000_f64).round() as u32;
+                let difference = Utc.timestamp(secs, nanosecs) - timestamp;
+                difference.num_nanoseconds().unwrap_or(0)
+            };
+
+            h.push(History {
+                id: uuid_v4(),
+                timestamp,
+                duration,
+                exit: entry.exit_code,
+                command: entry.cmd_line,
+                cwd: entry.pwd,
+                session: uuid_v4(),
+                hostname: entry.host,
+            })
+            .await?;
         }
 
-        // .resh_history.json lies about being a json. It is actually a file containing valid json
+        Ok(())
-        // on every line. This means that the last line will throw an error, as it is just an EOF.
-        // Without the special case here, that will crash the importer.
-        let entry = match serde_json::from_str::<ReshEntry>(&self.strbuf) {
-            Ok(e) => e,
-            Err(e) if e.is_eof() => return None,
-            Err(e) => {
-                return Some(Err(eyre!(
-                    "Invalid entry found in resh_history file: {}",
-                    e
-                )))
-            }
-        };
-
-        #[allow(clippy::cast_possible_truncation)]
-        #[allow(clippy::cast_sign_loss)]
-        let timestamp = {
-            let secs = entry.realtime_before.floor() as i64;
-            let nanosecs = (entry.realtime_before.fract() * 1_000_000_000_f64).round() as u32;
-            Utc.timestamp(secs, nanosecs)
-        };
-        #[allow(clippy::cast_possible_truncation)]
-        #[allow(clippy::cast_sign_loss)]
-        let duration = {
-            let secs = entry.realtime_after.floor() as i64;
-            let nanosecs = (entry.realtime_after.fract() * 1_000_000_000_f64).round() as u32;
-            let difference = Utc.timestamp(secs, nanosecs) - timestamp;
-            difference.num_nanoseconds().unwrap_or(0)
-        };
-
-        Some(Ok(History {
-            id: uuid_v4(),
-            timestamp,
-            duration,
-            exit: entry.exit_code,
-            command: entry.cmd_line,
-            cwd: entry.pwd,
-            session: uuid_v4(),
-            hostname: entry.host,
-        }))
-    }
-
-    fn size_hint(&self) -> (usize, Option<usize>) {
-        (self.loc, Some(self.loc))
     }
 }

atuin-client/src/import/zsh.rs

@@ -1,138 +1,104 @@
 // import old shell history!
 // automatically hoover up all that we can find
 
-use std::{
+use std::{fs::File, io::Read, path::PathBuf};
-    fs::File,
-    io::{BufRead, BufReader, Read, Seek},
-    path::{Path, PathBuf},
-};
 
+use async_trait::async_trait;
 use chrono::{prelude::*, Utc};
 use directories::UserDirs;
 use eyre::{eyre, Result};
-use itertools::Itertools;
 
-use super::{count_lines, Importer};
+use super::{get_histpath, unix_byte_lines, Importer, Loader};
 use crate::history::History;
 
 #[derive(Debug)]
-pub struct Zsh<R> {
+pub struct Zsh {
-    file: BufReader<R>,
+    bytes: Vec<u8>,
-    strbuf: String,
-    loc: usize,
-    counter: i64,
 }
 
-impl<R: Read + Seek> Zsh<R> {
+fn default_histpath() -> Result<PathBuf> {
-    fn new(r: R) -> Result<Self> {
+    // oh-my-zsh sets HISTFILE=~/.zhistory
-        let mut buf = BufReader::new(r);
+    // zsh has no default value for this var, but uses ~/.zhistory.
-        let loc = count_lines(&mut buf)?;
+    // we could maybe be smarter about this in the future :)
+    let user_dirs = UserDirs::new().ok_or_else(|| eyre!("could not find user directories"))?;
+    let home_dir = user_dirs.home_dir();
 
-        Ok(Self {
+    let mut candidates = [".zhistory", ".zsh_history"].iter();
-            file: buf,
+    loop {
-            strbuf: String::new(),
+        match candidates.next() {
-            loc,
+            Some(candidate) => {
-            counter: 0,
+                let histpath = home_dir.join(candidate);
-        })
+                if histpath.exists() {
-    }
+                    break Ok(histpath);
-}
-
-impl Importer for Zsh<File> {
-    const NAME: &'static str = "zsh";
-
-    fn histpath() -> Result<PathBuf> {
-        // oh-my-zsh sets HISTFILE=~/.zhistory
-        // zsh has no default value for this var, but uses ~/.zhistory.
-        // we could maybe be smarter about this in the future :)
-        let user_dirs = UserDirs::new().unwrap();
-        let home_dir = user_dirs.home_dir();
-
-        let mut candidates = [".zhistory", ".zsh_history"].iter();
-        loop {
-            match candidates.next() {
-                Some(candidate) => {
-                    let histpath = home_dir.join(candidate);
-                    if histpath.exists() {
-                        break Ok(histpath);
-                    }
                 }
-                None => break Err(eyre!("Could not find history file. Try setting $HISTFILE")),
             }
+            None => break Err(eyre!("Could not find history file. Try setting $HISTFILE")),
         }
     }
-
-    fn parse(path: impl AsRef<Path>) -> Result<Self> {
-        Self::new(File::open(path)?)
-    }
 }
 
-impl<R: Read> Iterator for Zsh<R> {
+#[async_trait]
-    type Item = Result<History>;
+impl Importer for Zsh {
+    const NAME: &'static str = "bash";
 
-    fn next(&mut self) -> Option<Self::Item> {
+    async fn new() -> Result<Self> {
-        // ZSH extended history records the timestamp + command duration
+        let mut bytes = Vec::new();
-        // These lines begin with :
+        let path = get_histpath(default_histpath)?;
-        // So, if the line begins with :, parse it. Otherwise it's just
+        let mut f = File::open(path)?;
-        // the command
+        f.read_to_end(&mut bytes)?;
-        self.strbuf.clear();
+        Ok(Self { bytes })
-        match self.file.read_line(&mut self.strbuf) {
+    }
-            Ok(0) => return None,
-            Ok(_) => (),
-            Err(e) => return Some(Err(eyre!("failed to read line: {}", e))), // we can skip past things like invalid utf8
-        }
 
-        self.loc -= 1;
+    async fn entries(&mut self) -> Result<usize> {
+        Ok(super::count_lines(&self.bytes))
+    }
 
-        while self.strbuf.ends_with("\\\n") {
+    async fn load(self, h: &mut impl Loader) -> Result<()> {
-            if self.file.read_line(&mut self.strbuf).is_err() {
+        let now = chrono::Utc::now();
-                // There's a chance that the last line of a command has invalid
+        let mut line = String::new();
-                // characters, the only safe thing to do is break :/
+
-                // usually just invalid utf8 or smth
+        let mut counter = 0;
-                // however, we really need to avoid missing history, so it's
+        for b in unix_byte_lines(&self.bytes) {
-                // better to have some items that should have been part of
+            let s = match std::str::from_utf8(b) {
-                // something else, than to miss things. So break.
+                Ok(s) => s,
-                break;
+                Err(_) => continue, // we can skip past things like invalid utf8
             };
 
-            self.loc -= 1;
+            if let Some(s) = s.strip_suffix('\\') {
+                line.push_str(s);
+                line.push_str("\\\n");
+            } else {
+                line.push_str(s);
+                let command = std::mem::take(&mut line);
+
+                if let Some(command) = command.strip_prefix(": ") {
+                    counter += 1;
+                    h.push(parse_extended(command, counter)).await?;
+                } else {
+                    let offset = chrono::Duration::seconds(counter);
+                    counter += 1;
+
+                    h.push(History::new(
+                        now - offset, // preserve ordering
+                        command.trim_end().to_string(),
+                        String::from("unknown"),
+                        -1,
+                        -1,
+                        None,
+                        None,
+                    ))
+                    .await?;
+                }
+            }
         }
 
-        // We have to handle the case where a line has escaped newlines.
+        Ok(())
-        // Keep reading until we have a non-escaped newline
-
-        let extended = self.strbuf.starts_with(':');
-
-        if extended {
-            self.counter += 1;
-            Some(Ok(parse_extended(&self.strbuf, self.counter)))
-        } else {
-            let time = chrono::Utc::now();
-            let offset = chrono::Duration::seconds(self.counter);
-            let time = time - offset;
-
-            self.counter += 1;
-
-            Some(Ok(History::new(
-                time,
-                self.strbuf.trim_end().to_string(),
-                String::from("unknown"),
-                -1,
-                -1,
-                None,
-                None,
-            )))
-        }
-    }
-
-    fn size_hint(&self) -> (usize, Option<usize>) {
-        (0, Some(self.loc))
     }
 }
 
 fn parse_extended(line: &str, counter: i64) -> History {
-    let line = line.replacen(": ", "", 2);
+    let (time, duration) = line.split_once(':').unwrap();
-    let (time, duration) = line.splitn(2, ':').collect_tuple().unwrap();
+    let (duration, command) = duration.split_once(';').unwrap();
-    let (duration, command) = duration.splitn(2, ';').collect_tuple().unwrap();
 
     let time = time
         .parse::<i64>()
@@ -158,64 +124,64 @@ fn parse_extended(line: &str, counter: i64) -> History {
 
 #[cfg(test)]
 mod test {
-    use std::io::Cursor;
-
     use chrono::prelude::*;
     use chrono::Utc;
+    use itertools::assert_equal;
+
+    use crate::import::tests::TestLoader;
 
     use super::*;
 
     #[test]
     fn test_parse_extended_simple() {
-        let parsed = parse_extended(": 1613322469:0;cargo install atuin", 0);
+        let parsed = parse_extended("1613322469:0;cargo install atuin", 0);
 
         assert_eq!(parsed.command, "cargo install atuin");
         assert_eq!(parsed.duration, 0);
         assert_eq!(parsed.timestamp, Utc.timestamp(1_613_322_469, 0));
 
-        let parsed = parse_extended(": 1613322469:10;cargo install atuin;cargo update", 0);
+        let parsed = parse_extended("1613322469:10;cargo install atuin;cargo update", 0);
 
         assert_eq!(parsed.command, "cargo install atuin;cargo update");
         assert_eq!(parsed.duration, 10_000_000_000);
         assert_eq!(parsed.timestamp, Utc.timestamp(1_613_322_469, 0));
 
-        let parsed = parse_extended(": 1613322469:10;cargo :b̷i̶t̴r̵o̴t̴ ̵i̷s̴ ̷r̶e̵a̸l̷", 0);
+        let parsed = parse_extended("1613322469:10;cargo :b̷i̶t̴r̵o̴t̴ ̵i̷s̴ ̷r̶e̵a̸l̷", 0);
 
         assert_eq!(parsed.command, "cargo :b̷i̶t̴r̵o̴t̴ ̵i̷s̴ ̷r̶e̵a̸l̷");
         assert_eq!(parsed.duration, 10_000_000_000);
         assert_eq!(parsed.timestamp, Utc.timestamp(1_613_322_469, 0));
 
-        let parsed = parse_extended(": 1613322469:10;cargo install \\n atuin\n", 0);
+        let parsed = parse_extended("1613322469:10;cargo install \\n atuin\n", 0);
 
         assert_eq!(parsed.command, "cargo install \\n atuin");
         assert_eq!(parsed.duration, 10_000_000_000);
         assert_eq!(parsed.timestamp, Utc.timestamp(1_613_322_469, 0));
     }
 
-    #[test]
+    #[tokio::test]
-    fn test_parse_file() {
+    async fn test_parse_file() {
-        let input = r": 1613322469:0;cargo install atuin
+        let bytes = r": 1613322469:0;cargo install atuin
 : 1613322469:10;cargo install atuin; \
 cargo update
 : 1613322469:10;cargo :b̷i̶t̴r̵o̴t̴ ̵i̷s̴ ̷r̶e̵a̸l̷
-";
+"
+        .as_bytes()
+        .to_owned();
 
-        let cursor = Cursor::new(input);
+        let mut zsh = Zsh { bytes };
-        let mut zsh = Zsh::new(cursor).unwrap();
+        assert_eq!(zsh.entries().await.unwrap(), 4);
-        assert_eq!(zsh.loc, 4);
-        assert_eq!(zsh.size_hint(), (0, Some(4)));
 
-        assert_eq!(&zsh.next().unwrap().unwrap().command, "cargo install atuin");
+        let mut loader = TestLoader::default();
-        assert_eq!(
+        zsh.load(&mut loader).await.unwrap();
-            &zsh.next().unwrap().unwrap().command,
+
-            "cargo install atuin; \\\ncargo update"
+        assert_equal(
+            loader.buf.iter().map(|h| h.command.as_str()),
+            [
+                "cargo install atuin",
+                "cargo install atuin; \\\ncargo update",
+                "cargo :b̷i̶t̴r̵o̴t̴ ̵i̷s̴ ̷r̶e̵a̸l̷",
+            ],
         );
-        assert_eq!(
-            &zsh.next().unwrap().unwrap().command,
-            "cargo :b̷i̶t̴r̵o̴t̴ ̵i̷s̴ ̷r̶e̵a̸l̷"
-        );
-        assert!(zsh.next().is_none());
-
-        assert_eq!(zsh.size_hint(), (0, Some(0)));
     }
 }

src/command/client.rs

@@ -58,6 +58,7 @@ pub enum Cmd {
 }
 
 impl Cmd {
+    #[tokio::main(flavor = "current_thread")]
     pub async fn run(self) -> Result<()> {
         pretty_env_logger::init();
 

src/command/client/history.rs

@@ -128,11 +128,7 @@ pub fn print_cmd_only(w: &mut StdoutLock, h: &[History]) {
 }
 
 impl Cmd {
-    pub async fn run(
+    pub async fn run(&self, settings: &Settings, db: &mut impl Database) -> Result<()> {
-        &self,
-        settings: &Settings,
-        db: &mut (impl Database + Send + Sync),
-    ) -> Result<()> {
         let context = current_context();
 
         match self {

src/command/client/import.rs

@@ -1,13 +1,14 @@
-use std::{env, path::PathBuf};
+use std::env;
 
+use async_trait::async_trait;
 use clap::Parser;
-use eyre::{eyre, Result};
+use eyre::Result;
 use indicatif::ProgressBar;
 
 use atuin_client::{
     database::Database,
     history::History,
-    import::{bash::Bash, fish::Fish, resh::Resh, zsh::Zsh, Importer},
+    import::{bash::Bash, fish::Fish, resh::Resh, zsh::Zsh, Importer, Loader},
 };
 
 #[derive(Parser)]
@@ -18,13 +19,10 @@ pub enum Cmd {
 
     /// Import history from the zsh history file
     Zsh,
-
     /// Import history from the bash history file
     Bash,
-
     /// Import history from the resh history file
     Resh,
-
     /// Import history from the fish history file
     Fish,
 }
@@ -32,7 +30,7 @@ pub enum Cmd {
 const BATCH_SIZE: usize = 100;
 
 impl Cmd {
-    pub async fn run(&self, db: &mut (impl Database + Send + Sync)) -> Result<()> {
+    pub async fn run<DB: Database>(&self, db: &mut DB) -> Result<()> {
         println!("        Atuin         ");
         println!("======================");
         println!("          \u{1f30d}          ");
@@ -47,124 +45,73 @@ impl Cmd {
 
                 if shell.ends_with("/zsh") {
                     println!("Detected ZSH");
-                    import::<Zsh<_>, _>(db, BATCH_SIZE).await
+                    import::<Zsh, DB>(db).await
                 } else if shell.ends_with("/fish") {
                     println!("Detected Fish");
-                    import::<Fish<_>, _>(db, BATCH_SIZE).await
+                    import::<Fish, DB>(db).await
                 } else if shell.ends_with("/bash") {
                     println!("Detected Bash");
-                    import::<Bash<_>, _>(db, BATCH_SIZE).await
+                    import::<Bash, DB>(db).await
                 } else {
                     println!("cannot import {} history", shell);
                     Ok(())
                 }
             }
 
-            Self::Zsh => import::<Zsh<_>, _>(db, BATCH_SIZE).await,
+            Self::Zsh => import::<Zsh, DB>(db).await,
-            Self::Bash => import::<Bash<_>, _>(db, BATCH_SIZE).await,
+            Self::Bash => import::<Bash, DB>(db).await,
-            Self::Resh => import::<Resh, _>(db, BATCH_SIZE).await,
+            Self::Resh => import::<Resh, DB>(db).await,
-            Self::Fish => import::<Fish<_>, _>(db, BATCH_SIZE).await,
+            Self::Fish => import::<Fish, DB>(db).await,
         }
     }
 }
 
-async fn import<I: Importer + Send, DB: Database + Send + Sync>(
+pub struct HistoryImporter<'db, DB: Database> {
-    db: &mut DB,
+    pb: ProgressBar,
-    buf_size: usize,
+    buf: Vec<History>,
-) -> Result<()>
+    db: &'db mut DB,
-where
+}
-    I::IntoIter: Send,
+
-{
+impl<'db, DB: Database> HistoryImporter<'db, DB> {
+    fn new(db: &'db mut DB, len: usize) -> Self {
+        Self {
+            pb: ProgressBar::new(len as u64),
+            buf: Vec::with_capacity(BATCH_SIZE),
+            db,
+        }
+    }
+
+    async fn flush(self) -> Result<()> {
+        if !self.buf.is_empty() {
+            self.db.save_bulk(&self.buf).await?;
+        }
+        self.pb.finish();
+        Ok(())
+    }
+}
+
+#[async_trait]
+impl<'db, DB: Database> Loader for HistoryImporter<'db, DB> {
+    async fn push(&mut self, hist: History) -> Result<()> {
+        self.pb.inc(1);
+        self.buf.push(hist);
+        if self.buf.len() == self.buf.capacity() {
+            self.db.save_bulk(&self.buf).await?;
+            self.buf.clear();
+        }
+        Ok(())
+    }
+}
+
+async fn import<I: Importer + Send, DB: Database>(db: &mut DB) -> Result<()> {
     println!("Importing history from {}", I::NAME);
 
-    let histpath = get_histpath::<I>()?;
+    let mut importer = I::new().await?;
-    let contents = I::parse(histpath)?;
+    let len = importer.entries().await.unwrap();
-
+    let mut loader = HistoryImporter::new(db, len);
-    let iter = contents.into_iter();
+    importer.load(&mut loader).await?;
-    let progress = if let (_, Some(upper_bound)) = iter.size_hint() {
+    loader.flush().await?;
-        ProgressBar::new(upper_bound as u64)
-    } else {
-        ProgressBar::new_spinner()
-    };
-
-    let mut buf = Vec::<History>::with_capacity(buf_size);
-    let mut iter = progress.wrap_iter(iter);
-    loop {
-        // fill until either no more entries
-        // or until the buffer is full
-        let done = fill_buf(&mut buf, &mut iter);
-
-        // flush
-        db.save_bulk(&buf).await?;
-
-        if done {
-            break;
-        }
-    }
 
     println!("Import complete!");
-
     Ok(())
 }
-
-fn get_histpath<I: Importer>() -> Result<PathBuf> {
-    if let Ok(p) = env::var("HISTFILE") {
-        is_file(PathBuf::from(p))
-    } else {
-        is_file(I::histpath()?)
-    }
-}
-
-fn is_file(p: PathBuf) -> Result<PathBuf> {
-    if p.is_file() {
-        Ok(p)
-    } else {
-        Err(eyre!(
-            "Could not find history file {:?}. Try setting $HISTFILE",
-            p
-        ))
-    }
-}
-
-fn fill_buf<T, E>(buf: &mut Vec<T>, iter: &mut impl Iterator<Item = Result<T, E>>) -> bool {
-    buf.clear();
-    loop {
-        match iter.next() {
-            Some(Ok(t)) => buf.push(t),
-            Some(Err(_)) => (),
-            None => break true,
-        }
-
-        if buf.len() == buf.capacity() {
-            break false;
-        }
-    }
-}
-
-#[cfg(test)]
-mod tests {
-    use super::fill_buf;
-
-    #[test]
-    fn test_fill_buf() {
-        let mut buf = Vec::with_capacity(4);
-        let mut iter = vec![
-            Ok(1),
-            Err(2),
-            Ok(3),
-            Ok(4),
-            Err(5),
-            Ok(6),
-            Ok(7),
-            Err(8),
-            Ok(9),
-        ]
-        .into_iter();
-
-        assert!(!fill_buf(&mut buf, &mut iter));
-        assert_eq!(buf, vec![1, 3, 4, 6]);
-
-        assert!(fill_buf(&mut buf, &mut iter));
-        assert_eq!(buf, vec![7, 9]);
-    }
-}

src/command/client/search.rs

@@ -75,11 +75,7 @@ pub struct Cmd {
 }
 
 impl Cmd {
-    pub async fn run(
+    pub async fn run(self, db: &mut impl Database, settings: &Settings) -> Result<()> {
-        self,
-        db: &mut (impl Database + Send + Sync),
-        settings: &Settings,
-    ) -> Result<()> {
         if self.interactive {
             let item = select_history(
                 &self.query,
@@ -257,7 +253,7 @@ impl State {
 async fn query_results(
     app: &mut State,
     search_mode: SearchMode,
-    db: &mut (impl Database + Send + Sync),
+    db: &mut impl Database,
 ) -> Result<()> {
     let results = match app.input.as_str() {
         "" => {
@@ -284,7 +280,7 @@ async fn query_results(
 async fn key_handler(
     input: Key,
     search_mode: SearchMode,
-    db: &mut (impl Database + Send + Sync),
+    db: &mut impl Database,
     app: &mut State,
 ) -> Option<String> {
     match input {
@@ -537,7 +533,7 @@ async fn select_history(
     search_mode: SearchMode,
     filter_mode: FilterMode,
     style: atuin_client::settings::Style,
-    db: &mut (impl Database + Send + Sync),
+    db: &mut impl Database,
 ) -> Result<String> {
     let stdout = stdout().into_raw_mode()?;
     let stdout = MouseTerminal::from(stdout);
@@ -596,7 +592,7 @@ async fn run_non_interactive(
     after: Option<String>,
     limit: Option<i64>,
     query: &[String],
-    db: &mut (impl Database + Send + Sync),
+    db: &mut impl Database,
 ) -> Result<()> {
     let dir = if cwd.as_deref() == Some(".") {
         let current = std::env::current_dir()?;

src/command/client/stats.rs

@@ -62,11 +62,7 @@ fn compute_stats(history: &[History]) -> Result<()> {
 }
 
 impl Cmd {
-    pub async fn run(
+    pub async fn run(&self, db: &mut impl Database, settings: &Settings) -> Result<()> {
-        &self,
-        db: &mut (impl Database + Send + Sync),
-        settings: &Settings,
-    ) -> Result<()> {
         let context = current_context();
         let words = if self.period.is_empty() {
             String::from("all")

src/command/client/sync.rs

@@ -31,11 +31,7 @@ pub enum Cmd {
 }
 
 impl Cmd {
-    pub async fn run(
+    pub async fn run(self, settings: Settings, db: &mut impl Database) -> Result<()> {
-        self,
-        settings: Settings,
-        db: &mut (impl Database + Send + Sync),
-    ) -> Result<()> {
         match self {
             Self::Sync { force } => run(&settings, force, db).await,
             Self::Login(l) => l.run(&settings).await,
@@ -52,11 +48,7 @@ impl Cmd {
     }
 }
 
-async fn run(
+async fn run(settings: &Settings, force: bool, db: &mut impl Database) -> Result<()> {
-    settings: &Settings,
-    force: bool,
-    db: &mut (impl Database + Send + Sync),
-) -> Result<()> {
     atuin_client::sync::sync(settings, force, db).await?;
     println!(
         "Sync complete! {} items in database, force: {}",

src/command/mod.rs

@@ -19,11 +19,11 @@ pub enum AtuinCmd {
 }
 
 impl AtuinCmd {
-    pub async fn run(self) -> Result<()> {
+    pub fn run(self) -> Result<()> {
         match self {
-            Self::Client(client) => client.run().await,
+            Self::Client(client) => client.run(),
             #[cfg(feature = "server")]
-            Self::Server(server) => server.run().await,
+            Self::Server(server) => server.run(),
         }
     }
 }

src/command/server.rs

@@ -21,6 +21,7 @@ pub enum Cmd {
 }
 
 impl Cmd {
+    #[tokio::main]
     pub async fn run(self) -> Result<()> {
         tracing_subscriber::registry()
             .with(fmt::layer())

src/main.rs

@@ -25,12 +25,11 @@ struct Atuin {
 }
 
 impl Atuin {
-    async fn run(self) -> Result<()> {
+    fn run(self) -> Result<()> {
-        self.atuin.run().await
+        self.atuin.run()
     }
 }
 
-#[tokio::main]
+fn main() -> Result<()> {
-async fn main() -> Result<()> {
+    Atuin::parse().run()
-    Atuin::parse().run().await
 }