ansible - example of using map + regex_replace to modify all items in a list

00_play.yml

---
- hosts: localhost
  connection: local
  gather_facts: False
  vars:
    mylist:
      - group_1
      - group_2

  tasks:
    - debug:
        msg: "{{ mylist | map('regex_replace', '(.*)', '\\1_foo') | list }}"

01_output

PLAY [localhost] **********************************************************************************************************************

TASK [debug] **************************************************************************************************************************
ok: [localhost] => {
    "changed": false,
    "msg": [
        "group_1_foo",
        "group_2_foo"
    ]
}

PLAY RECAP ****************************************************************************************************************************
localhost                  : ok=1    changed=0    unreachable=0    failed=0

The \ doesn't appear to make a difference. I tried '\', but that didn't happen.

Another approach should be:

    - debug: msg="YYY {{ groups['all'] | map('regex_replace', '(.*)', "hostvars[\1].ansible_ssh_host") | list }}"

which correctly outputs:

ok: [neptune] => {
    "changed": false, 
    "msg": "YYY [u'hostvars[unifi-controller-office].ansible_ssh_host', u'hostvars[neptune].ansible_ssh_host']"
}

So the question is, how do we evaluate those strings to get them replaced by the variable values.

this is a method for modifying all items in a list, it's not suitable for pulling out elements from a nested structure.

As pointed out in chan,

{{ groups['all'] | map('extract', hostvars, 'ansible_ssh_host' }}

is the way to do that

if you need to both add a prefix and a suffix, as well as making everything a list, take a look at the below:

  set_fact:
    extended_etcd_endpoints_list: "{{ groups['etcd'] | map('extract', hostvars, ['ansible_default_ipv4','address']) | map('regex_replace', '^(.*)$','https://\\1:2379') | list  }}"

it takes the list of all machines in the group etcd, extracts the ipv4, adds an 'https://' in front and an ':2379' at the end.
Finally, everything is transformed in a list.

Indeed, I got strange results if the regex was not achored at the beginning: "^..."
Other than that, thanks for this useful pattern.