Merge pull request #374 from TD-Sky/dump-history

feat(dump history): You can specify the time range, regex and format to dump

Cargo.toml

@@ -17,7 +17,10 @@ debug = true
 
 [dependencies]
 chrono = "0.4"
+chrono-systemd-time = "0.3"
 csv = "1"
+serde_json = "1"
+serde = { version = "1", features = ["derive"] }
 humantime = "2.1"
 directories-next = "2.0"
 itertools = "0.10"
@@ -36,7 +39,6 @@ features = ["functions", "unlock_notify"]
 version = "0.26"
 features = ["use-dev-tty"]
 
-
 [dependencies.clap]
 version = "4"
 features = ["derive"]

README.md

@@ -206,6 +206,59 @@ To avoid McFly's UI messing up your scrollback history in iTerm2, make sure this
 
 <img src="/docs/iterm2.jpeg" alt="iterm2 UI instructions">
 
+## Dump history
+
+McFly can dump the command history into *stdout*.
+
+For example:
+```bash
+mcfly dump --since '2023-01-01' --before '2023-09-12 09:15:30'
+```
+will dump the command run between *2023-01-01 00:00:00.0* to *2023-09-12 09:15:30*(**exclusive**) as **json**.
+You can specify **csv** as dump format via `--format csv` as well.
+
+Each item in dumped commands has the following fields:
+* `cmd`: The run command.
+* `when_run`: The time when the command ran in your local timezone.
+
+You can dump all the commands history without any arguments:
+```bash
+mcfly dump
+```
+
+### Timestamp format
+McFly use [chrono-systemd-time-ng] parsing timestamp.
+
+**chrono-systemd-time-ng** is a non-strict implementation of [systemd.time](https://www.freedesktop.org/software/systemd/man/systemd.time.html), with the following exceptions:
+* time units **must** accompany all time span values.
+* time zone suffixes are **not** supported.
+* weekday prefixes are **not** supported.
+
+Users of McFly simply need to understand **specifying timezone in timestamp isn't allowed**.
+McFly will always use your **local timezone**.
+
+For more details, please refer to [the document of chrono-systemd-time-ng][chrono-systemd-time-ng].
+
+[chrono-systemd-time-ng]: https://docs.rs/chrono-systemd-time-ng/latest/chrono_systemd_time/
+
+### Regex
+*Dump* supports filtering commands with regex.
+The regex syntax follows [crate regex](https://docs.rs/regex/latest/regex/#syntax).
+
+For example:
+```bash
+mcfly dump -r '^cargo run'
+```
+will dump all command prefixes with `cargo run`.
+
+You can use `-r/--regex` and time options at the same time.
+
+For example:
+```bash
+mcfly dump -r '^cargo run' --since '2023-09-12 09:15:30'
+```
+will dump all command prefixes with `cargo run` ran since *2023-09-12 09:15:30*.
+
 ## Settings
 A number of settings can be set via environment variables. To set a setting you should add the following snippets to your `~/.bashrc` / `~/.zshrc` / `~/.config/fish/config.fish`.
 

src/cli.rs

@@ -1,4 +1,5 @@
 use clap::{Parser, Subcommand, ValueEnum};
+use regex::Regex;
 use std::path::PathBuf;
 
 /// Fly through your shell history
@@ -108,6 +109,36 @@ pub enum SubCommand {
         #[arg(value_enum)]
         shell: InitMode,
     },
+
+    /// Dump history into stdout; the results are sorted by timestamp
+    Dump {
+        /// Select all commands ran since the point
+        #[arg(long)]
+        since: Option<String>,
+
+        /// Select all commands ran before the point
+        #[arg(long)]
+        before: Option<String>,
+
+        /// Sort order [case ignored]
+        #[arg(
+            long,
+            short,
+            value_name = "ORDER",
+            value_enum,
+            default_value_t,
+            ignore_case = true
+        )]
+        sort: SortOrder,
+
+        /// Require commands to match the pattern
+        #[arg(long, short)]
+        regex: Option<Regex>,
+
+        /// The format to dump in
+        #[arg(long, short, value_enum, default_value_t)]
+        format: DumpFormat,
+    },
 }
 
 #[derive(Clone, Copy, ValueEnum, Default)]
@@ -126,8 +157,35 @@ pub enum InitMode {
     Fish,
 }
 
+#[derive(Debug, Clone, Copy, ValueEnum, Default)]
+#[value(rename_all = "UPPER")]
+pub enum SortOrder {
+    #[default]
+    #[value(alias = "asc")]
+    Asc,
+    #[value(alias = "desc")]
+    Desc,
+}
+
+#[derive(Debug, Clone, Copy, ValueEnum, Default)]
+pub enum DumpFormat {
+    #[default]
+    Json,
+    Csv,
+}
+
 impl Cli {
     pub fn is_init(&self) -> bool {
         matches!(self.command, SubCommand::Init { .. })
     }
 }
+
+impl SortOrder {
+    #[inline]
+    pub fn to_str(&self) -> &'static str {
+        match self {
+            Self::Asc => "ASC",
+            Self::Desc => "DESC",
+        }
+    }
+}

src/dumper.rs

@@ -0,0 +1,56 @@
+use std::io::{self, BufWriter, Write};
+
+use crate::cli::DumpFormat;
+use crate::history::{DumpCommand, History};
+use crate::settings::Settings;
+use crate::time::to_datetime;
+
+#[derive(Debug)]
+pub struct Dumper<'a> {
+    settings: &'a Settings,
+    history: &'a History,
+}
+
+impl<'a> Dumper<'a> {
+    #[inline]
+    pub fn new(settings: &'a Settings, history: &'a History) -> Self {
+        Self { settings, history }
+    }
+
+    pub fn dump(&self) {
+        let mut commands = self
+            .history
+            .dump(&self.settings.time_range, &self.settings.sort_order);
+        if commands.is_empty() {
+            println!("McFly: No history");
+            return;
+        }
+
+        if let Some(pat) = &self.settings.pattern {
+            commands.retain(|dc| pat.is_match(&dc.cmd));
+        }
+
+        match self.settings.dump_format {
+            DumpFormat::Json => Self::dump2json(&commands),
+            DumpFormat::Csv => Self::dump2csv(&commands),
+        }
+        .unwrap_or_else(|err| panic!("McFly error: Failed while output history ({err})"));
+    }
+}
+
+impl<'a> Dumper<'a> {
+    fn dump2json(commands: &[DumpCommand]) -> io::Result<()> {
+        let mut stdout = BufWriter::new(io::stdout().lock());
+        serde_json::to_writer_pretty(&mut stdout, commands).map_err(io::Error::from)?;
+        stdout.flush()
+    }
+
+    fn dump2csv(commands: &[DumpCommand]) -> io::Result<()> {
+        let mut wtr = csv::Writer::from_writer(io::stdout().lock());
+        wtr.write_record(["cmd", "when_run"])?;
+        for dc in commands {
+            wtr.write_record([dc.cmd.as_str(), to_datetime(dc.when_run).as_str()])?;
+        }
+        wtr.flush()
+    }
+}

src/history/history.rs

@@ -1,20 +1,22 @@
 #![allow(clippy::module_inception)]
-use crate::shell_history;
-use rusqlite::named_params;
-use rusqlite::{Connection, MappedRows, Row};
-use std::cmp::Ordering;
-use std::io::Write;
-use std::path::PathBuf;
-use std::{fmt, fs, io};
-//use std::time::Instant;
+use crate::cli::SortOrder;
 use crate::history::{db_extensions, schema};
 use crate::network::Network;
 use crate::path_update_helpers;
-use crate::settings::{HistoryFormat, ResultFilter, ResultSort, Settings};
+use crate::settings::{HistoryFormat, ResultFilter, ResultSort, Settings, TimeRange};
+use crate::shell_history;
 use crate::simplified_command::SimplifiedCommand;
+use crate::time::to_datetime;
 use itertools::Itertools;
+use rusqlite::named_params;
 use rusqlite::types::ToSql;
+use rusqlite::{Connection, MappedRows, Row};
+use serde::{Serialize, Serializer};
+use std::cmp::Ordering;
+use std::io::Write;
+use std::path::PathBuf;
 use std::time::{Instant, SystemTime, UNIX_EPOCH};
+use std::{fmt, fs, io};
 
 #[derive(Debug, Clone, Default)]
 pub struct Features {
@@ -46,6 +48,13 @@ pub struct Command {
     pub match_bounds: Vec<(usize, usize)>,
 }
 
+#[derive(Debug, Clone, Serialize)]
+pub struct DumpCommand {
+    pub cmd: String,
+    #[serde(serialize_with = "ser_to_datetime")]
+    pub when_run: i64,
+}
+
 impl fmt::Display for Command {
     fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
         self.cmd.fmt(f)
@@ -58,6 +67,14 @@ impl From<Command> for String {
     }
 }
 
+#[inline]
+fn ser_to_datetime<S>(when_run: &i64, serializer: S) -> Result<S::Ok, S::Error>
+where
+    S: Serializer,
+{
+    serializer.serialize_str(&to_datetime(*when_run))
+}
+
 #[derive(Debug)]
 pub struct History {
     pub connection: Connection,
@@ -631,8 +648,22 @@ impl History {
             format!("SELECT id, cmd, cmd_tpl, session_id, when_run, exit_code, selected, dir FROM commands WHERE session_id = :session_id ORDER BY {} DESC LIMIT :limit OFFSET :offset", order)
         };
 
+        let closure: fn(&Row) -> rusqlite::Result<Command> = |row| {
+            Ok(Command {
+                id: row.get(0)?,
+                cmd: row.get(1)?,
+                cmd_tpl: row.get(2)?,
+                session_id: row.get(3)?,
+                when_run: row.get(4)?,
+                exit_code: row.get(5)?,
+                selected: row.get(6)?,
+                dir: row.get(7)?,
+                ..Command::default()
+            })
+        };
+
         if session_id.is_none() {
-            self.run_query(&query, &[(":limit", &num), (":offset", &offset)])
+            self.run_query(&query, &[(":limit", &num), (":offset", &offset)], closure)
         } else {
             self.run_query(
                 &query,
@@ -641,34 +672,24 @@ impl History {
                     (":limit", &num),
                     (":offset", &offset),
                 ],
+                closure,
             )
         }
     }
 
-    fn run_query(&self, query: &str, params: &[(&str, &dyn ToSql)]) -> Vec<Command> {
+    fn run_query<T, F>(&self, query: &str, params: &[(&str, &dyn ToSql)], f: F) -> Vec<T>
+    where
+        F: FnMut(&Row<'_>) -> rusqlite::Result<T>,
+    {
         let mut statement = self.connection.prepare(query).unwrap();
 
-        let closure: fn(&Row) -> Result<Command, _> = |row| {
-            Ok(Command {
-                id: row.get(0)?,
-                cmd: row.get(1)?,
-                cmd_tpl: row.get(2)?,
-                session_id: row.get(3)?,
-                when_run: row.get(4)?,
-                exit_code: row.get(5)?,
-                selected: row.get(6)?,
-                dir: row.get(7)?,
-                ..Command::default()
-            })
-        };
-
-        let command_iter: MappedRows<_> = statement
-            .query_map(params, closure)
+        let rows: MappedRows<_> = statement
+            .query_map(params, f)
             .unwrap_or_else(|err| panic!("McFly error: Query Map to work ({})", err));
 
-        let mut vec = Vec::new();
-        for command in command_iter.flatten() {
-            vec.push(command);
+        let mut vec: Vec<T> = Vec::new();
+        for row in rows.flatten() {
+            vec.push(row);
         }
 
         vec
@@ -755,6 +776,44 @@ impl History {
         }
     }
 
+    pub fn dump(&self, time_range: &TimeRange, order: &SortOrder) -> Vec<DumpCommand> {
+        let mut where_clause = String::new();
+        // Were there condtions in where clause?
+        let mut has_conds = false;
+        let mut params: Vec<(&str, &dyn ToSql)> = Vec::with_capacity(2);
+
+        if !time_range.is_full() {
+            where_clause.push_str("WHERE");
+
+            if let Some(since) = &time_range.since {
+                where_clause.push_str(" :since <= when_run");
+                has_conds = true;
+                params.push((":since", since));
+            }
+
+            if let Some(before) = &time_range.before {
+                if has_conds {
+                    where_clause.push_str(" AND");
+                }
+
+                where_clause.push_str(" when_run < :before");
+                params.push((":before", before));
+            }
+        }
+
+        let query = format!(
+            "SELECT cmd, when_run FROM commands {} ORDER BY when_run {}",
+            where_clause,
+            order.to_str()
+        );
+        self.run_query(&query, params.as_slice(), |row| {
+            Ok(DumpCommand {
+                cmd: row.get(0)?,
+                when_run: row.get(1)?,
+            })
+        })
+    }
+
     fn from_shell_history(history_format: HistoryFormat) -> History {
         print!(
             "McFly: Importing shell history for the first time. This may take a minute or two..."

src/history/mod.rs

@@ -1,4 +1,4 @@
-pub use self::history::{Command, Features, History};
+pub use self::history::{Command, DumpCommand, Features, History};
 
 mod db_extensions;
 mod history;