ah

ah is a complementary software for a builtin shell history command you've used to use for years and I hope you've dreamt about it as I did.

It is not a replacement for history but anyway it perfectly matches a common history | grep pattern of usage but it allows you to do a bit more. It allows you to trace an output of a command, to fetch it from the archive, to bookmark some commands and to execute them.

How often do you kick yourself for loosing important output of your SSH session? Sometimes you use screen or tmux for that purposes but it is pretty awkward to search through it. Or how often do you find yourself typing CTRL+R or juggling with Up and Down buttons to find something like make && mv -f coolapp $DIR/bin && coolapp. Stop it for a great good, ah will likely help you here.

Currently it supports following features:

Tracing an output
Fetching the output trace
Bookmarking command
Executing of a command by number or bookmark
Showing a history with greping on regular expression or fuzzy match

ah does not maintains its own history file, it uses your regular ~/.bash_history or .zsh_history. So no worries here: bash or zsh maintains a history and ah gives you several features on the top.

ah supports Zsh and Bash.

Installation

You may build ah from sources or just download proper binary from releases.

To install it from sources, just do following:

$ git clone https://github.com/9seconds/ah.git $GOPATH/src/github.com/9seconds/ah
$ cd $GOPATH/src/github.com/9seconds/ah
$ make install

It will copy the binary into $GOBIN/ah. Or you may do that:

$ git clone https://github.com/9seconds/ah.git $GOPATH/src/github.com/9seconds/ah
$ cd $GOPATH/src/github.com/9seconds/ah
$ make
$ mv ah /wherever/you/want

Also, if you use HomeBrew/LinuxBrew, you may want to check the formula:

$ brew tap 9seconds/homebrew-ah
$ brew install ah

Update with Brews are trivial

$ brew update
$ brew reinstall ah

Tracing an output

So you want ~~to be a hero~~ to capture an output of some of your commands. Usually if you know that output is rather important, you use tee command.

$ find . -name "*.go" -type f | tee files.log

The main problem here is that only stdout will go to the files.log. You will lose stderr, right? The common way of solving that is redirecting a streams

$ find . -name "*.go" -type f 2>&1 | tee files.log

$ find . -name "*.go" -type f |& tee files.log

in a recent Bash-compatible shells. The problem here, you are streams are dangerously mixed into one and there is no way to pipe them into different processes. Let's say, you want to have output stored persistently but in the same time you want to have only filtered log messages on the screen. You can't do something like this.

$ find . -name "*.go" -type f |& tee files.log > /dev/null 2> grep -i localhost

Okay, this a rare case but please notice that ah knows how to handle that. Let's talk on how to store this output, where to keep it. You may store it somehow but it is a way better to have a tool which allows you not to think on how to name this output and how to remember which was the command.

Let's talk about ah now. Ah has its main t command which works rather simple.

$ ah t -- find . -name "*.go" -type f

Thats all. You will see output on the screen and you may pipe both streams wherever you want! Ah will store it persistently. And it will finish execution with precisely the same exit code as the original command does. Neat, right?

If you want to run a program which requires a pseudo TTY, just use -y option. And if you want to have your aliases to work, just run it with -x option!

Ah supports SSH and you may even run curses apps there, they will work, no worries.

Show the history

Now let's talk about viewing the history. ah does that with s command.

$ ah s
...
!10024  (01.11.14 16:01:09) *  ah t -- find . -name "*.go" -type f

What do we have here: we have banged command number (gues why it has ! here), we have a date (yes, HISTTIMEFORMAT supported!), we have a rather strange star mark and a command. What does that star mark mean? Basically it just shows that ah keeps a mixed output of that command and you may fetch it on demand.

ah has -g options which allows you to grep this list. Argument - a regular expression. It also has a convenient flag -z which activates fuzzy match. It works like this

$ ah s -z -g doigreREPOsoru

And I will see matched docker images | egrep -v 'REPOSITORY|<none>' | cut -d' ' -f1 | sort -u I bold important letters here. Basically I do it thinking like this: "I want do cker i mages, it was gre p REPO SITORY and sor ted with -u" typing just a few letters.

It also supports number argument. Let's say ah s 10 will show latest 10 commands, ah s 10 20 will show commands from 10 to 20. Also negative numbers are supported (but with underscore prefix, not hyphen), they are mostly work as Python slices. ah s 10 _20 means literally "from 10 to the latest 20". Basically ah s 10 equal to ah s _10 _1

Show an output

Output could be checked with l command. Just type ah l 10024 and you are good.

Bookmarks

You may pin any command number with bookmark using b command. After that you may execute it with e command. To fetch a list of bookmarks use lb commands, to remove several, use rb command.

So simple.

Garbage collecting

If you do not need a lot of traces or bookmarks, you may get rid of them using gt (garbage collect traces) and gb (garbage collect bookmarks) commands.

Automatic execution

(Only for zsh)

Ah allows user to set it for automatic tracing outputs. To do that please do the following:

You have to download script to source and place it wherever you want. Let's say, I put it to ~/.auto_ah.sh
Add following line to your ~/.zshrc: source ~/.zshrc and please be noticed that ah should be in your PATH. If which ah works, then you're done. Otherwise just add it to the path.

Basically, ah will track all your executions automatically. But since it is dangerous to execute automatically everything around, there is a whitelist. ah has 3 commands you should be interested in:

al shows the list of all whitelist command ah will automatically apply to.
ar removes command from the whitelist and ah won't execute it automatically.
ad add a command to the whitelist.

Let's check my current setup.

$ ah al
ag                   [interactive=false, pseudoTty=false]
aptg                 [interactive=true , pseudoTty=false]
awk                  [interactive=false, pseudoTty=false]
docker               [interactive=false, pseudoTty=true ]
docker_clean         [interactive=true , pseudoTty=false]
docker_stop          [interactive=true , pseudoTty=false]
docker_update        [interactive=true , pseudoTty=false]
find                 [interactive=false, pseudoTty=false]
grep                 [interactive=false, pseudoTty=false]
ipython              [interactive=true , pseudoTty=true ]
make                 [interactive=false, pseudoTty=false]
python               [interactive=false, pseudoTty=false]
sed                  [interactive=false, pseudoTty=false]
ssh                  [interactive=false, pseudoTty=false]
vagrant              [interactive=true , pseudoTty=true ]

As you can see, I have a mixed setup. I trace an output of ag or find command and do it in non-interactive (interactive means zsh -i -c) way and do not allocate pseudo TTY for them. There are several aliases (docker_update or aptg) and to execute them I use interactive mode. And I use pseudo TTY for ipython.

Now let's add go for the list.

$ ah ad go

So simple. But how can I set interactiveness or pseudo TTYs? Pretty simple and obvious (remember t command?)

$ ah ad -x go

for interactiveness. And for pseudo TTY

$ ah ad -y go

If you decide to use another set of options, just execute ah ad with another set of options, it will override previous setting.

To remove command just use ar

$ ah ar go

No need to resource or do something more.

Configuration

ah supports configuration with YAML file. It should be placed in ~/.ah/config.yaml. Here is the full example (everything may be omit)

shell: zsh
histfile: /home/9seconds/.zsh_history
histtimeformat: "%d.%m.%y %H:%M:%S"

tmpdir: /tmp

That simple, yes. It is useful, if you bring a lot of commandline options in aliases or if you want to execute ah automatically.

Here is the sequence of argument overriding:

Default options
Config options
Commandline options

So commandline options overrides config.