dysosmus / ansible-completion Goto Github PK

According to this documentation page, in --limit option we can use a filename instead of host names. To do it we prepend it with "@" symbol.

In this line it should check if current word starts with "@" and do another compreply for that case.

There are a few more ansible tools. Namely:

• ansible-doc
• ansible-galaxy
• ansible-playbook
• ansible-pull
• ansible-vault

Would you like to implement completion for those as well in this repo or at least accept pull-requests?

It writen only for freebsd?
In linux md5->md5sum and other keys to stat.
It work on linux whith -

@@ -86 +86 @@
-    local hash_module_path=$(md5 -q -s "$module_path")
+    local hash_module_path=$(echo "$module_path" | md5sum |awk {'print $1'})
@@ -116 +117 @@
-    echo $(stat -f "%Sm" -t "%s" $1)
+    echo $(stat -c "%Z" $1)

In the readme the step 2 tells:

2. Copy/move the ansible-completion.bash in your bash_completion.d
   folder (/etc/bash_completion.d, /usr/local/etc/bash_completion.dor ~/bash_completion.d).

But that path does not exist anymore. See https://archlinux.org/todo/use-usrsharebash-completioncompletions-for-package-bash-completions/

Instead, I think each *-completion.bash file except ansible-completion.bash should source the common one. I.e. include the following line after shebang:
source /usr/share/bash-completion/completions/ansible

Looks like the file names for ansible-pull and ansible-vault got switched around.

Thanks!
Scott

Edit: oh, and the ansible-vault function is also labeled as _ansible-pull.

If enable -
set -o noclobber
it have errors like

ansible myhost -m bash: /tmp/25576.module-name.ansible.completion: cannot overwrite existing file

As variant it fixed -

-    local cache_file=/tmp/${$}.${module_path:+"$hash_module_path".}module-name.ansible.completion 
+    # /tmp/<pid>.<hash of the module path if exsist>.module-name.ansible.completion and time for set -o noclobber
+    local cache_file=/tmp/${$}.$(date +%s).${module_path:+"$hash_module_path".}module-name.ansible.completion 

And it work, but it doesnt good solution, i think -

find /tmp/ -name "*.completion" 
/tmp/25576.1399165774.module-name.ansible.completion
/tmp/25576.1399165789.module-name.ansible.completion
/tmp/25576.1399165767.module-name.ansible.completion
/tmp/25576.module-name.ansible.completion
/tmp/25576.1399165783.module-name.ansible.completion
/tmp/25576.1399165770.module-name.ansible.completion

I wanted to point out that the installation steps for a user-defined bash completion

A) direct users to make a non hidden directory for thier user configuration, which breaks the norm and
B) doesn't actually get autoloaded by the bash shell.

However, according to http://serverfault.com/q/506612/210994

a user can make a file ~/.bash_completion and append to that file with the completions they want.

Or even better, make a directory ~/.bash_completion.d/ and populate the ~/.bash_completion file with the body:

~/.bash_completion
#!/usr/bin/env bash

if [ -d "$HOME/.bash_completion.d" ]; then
    for filename in "$HOME/.bash_completion.d/*"; do
        source $filename
    done
fi

Any file placed in ~/.bash_completion.d/ will be loaded.

This utilizes the existing bash autoload, And leaves the config files as hidden files to follow precedent.

Hello. Thanks for this project.
Seems like this project is completely separate and unofficial. Could be this mentioned in the readme?

Also, how it differs from the python-argcomplete method, described in official documentation here? Could this be mentioned in readme too?

Hi,
I wanted to learn about bash completion by writing something like this, for ansible-playbook.
In particular, I wanted to make it so, that when you provide a parameter for the inventory, for example -i inventory/my-inventory and then you add the --limit parameter, it would complete the possible parameter values according to what is in that inventory.
Would you consider that a suitable addition?
Not sure how hard it is though... ;)