rash 0.1.0

What is this?

Shell history is useful. But it can be more useful if it logs more data points. For example, if you forget which make target to run for certain project, you’d want to search shell commands that are ran in particular directory. Wouldn’t it be nice if you can do this?:

rash search --cwd . "make*"

RASH records many data points and they are stored in SQLite database. Here is a list of recorded information [1].

1. Current directory ($PWD).

2. Exit code ($?)

3. Exit code of pipes ($PIPESTATUS / $pipestatus)

4. The time command is started and terminated.

5. Environment variable ($PATH, $SHELL, $TERM, $HOST, etc.)

6. Real terminal. $TERM is used to fake programs. RASH can detect if you are in tmux, byobu, screen, gnome-terminal, etc.

7. Session information. If you go back and forth in some terminals, RASH does not loose in which sequence you ran the commands in which terminal.

Install

RASH is written in Python. The easiest way to install is to use pip (or easy_install, if you wish). You may need sudo for installing it in a system directory.:

pip install rash

RASH tested against Python 2.6, 2.7 and 3.2. However, as watchdog does not work with Python 3, you can’t get full power of RASH with Python 3.

If you want to use developmental version, just clone the git repository and add the following in your RC file.:

alias rash="PATH/TO/RASH/rash_cli.py"

Setup

Add this to your .zshrc or .bashrc. That’s all.:

eval "$(rash init)"

For more information, see rash init --help.

Usage

rash search --cwd . "*test*" "tox*"

rash search --time-after "1 week ago" "git*"

rash search --cwd-under . --include-exit-code 0 --duration-longer-than 30m

rash search --after-context 5 "cd SOME-DIRECTORY"

rash search --cwd . --exclude-exit-code 0

rash search --limit 5 --sort-by-program-frequency

rash search --limit -1 --no-unique --time-after "1 day ago" | wc -l

% rash search --with-command-id --limit 5 "*git*"
 359  git log
1253  git help clone
1677  git help diff
1678  git diff --word-diff
1780  git merge

rash show 1677

# Type `Ctrl-x r` to start isearch
bindkey "^Xr" rash-zle-isearch

rash-isearch(){
  eval "$(rash isearch)"
}

Dependency

Python modules:

• watchdog [2]

• parsedatetime [2]

Design principle

RASH’s design is focused on sparseness. There are several stages of data transformation until you see the search result, and they are done by separated processes.

First, rash record command dumps shell history in raw JSON record. This part of program does not touches to DB to make process very fast. As there is no complex transformation in this command, probably in the future version is is better to rewrite it entirely in shell function.

Second, rash daemon runs in background and watches the directory to store JSON record. When JSON record arrives, it insert the data into database.

rash record and rash daemon are setup by simple shell snippet eval $(rash init).

Finally, you can search through command history using search interface such as rash search. This search is very fast as you don’t read all JSON records in separated files.

+-------+         +--------+         +--------+         +--------+
| Shell |         | Raw    |         | SQLite |         | Search |
| hooks |-------->| JSON   |-------->|   DB   |-------->| result |
+-------+         | record |         +--------+         +--------+
                  +--------+

        `rash record`      `rash daemon`      `rash search`
                                               `rash show`

        \------------------------------/      \------------/
           `rash init` setups them           search interface

License

RASH is licensed under GPL v3. See COPYING for details.