Ansible - problem with ansible_facts for packages in jinja template - ansible

Question

I try to get all installed packages with Ansible and write them in a "pretty" way to a file.
Calling the module works:
- name: Gather the rpm package facts
package_facts:
manager: auto
In a Jinja template I am using a loop, what works too:
{% for item in ansible_facts.packages %}
{{ item }}
{% endfor %}
Unfortunately the simple output creates this "mess":
"yum": [
{
"arch": "noarch",
"epoch": null,
"name": "yum",
"release": "4.el8",
"source": "rpm",
"version": "4.2.23"
}
],
"zlib": [
{
"arch": "x86_64",
"epoch": null,
"name": "zlib",
"release": "16.el8_2",
"source": "rpm",
"version": "1.2.11"
}
]
Some of these elements are unnecassry for the current job, so the first call coming in mind was this:
{% for item in ansible_facts.packages %}
{{ item.name }} {{ item.version }}
{% endfor %}
But this ended just in an error:
fatal: [somehost.example.org]: FAILED! => {"changed": false, "msg": "AnsibleUndefinedVariable: 'unicode object' has no attribute 'name'"}
Searched through the internet, looked into the documentation of Ansible, tried various notations and nothing worked:
vars[item].name
item[0].name
item["name"]
As last option I tried it with iteritems:
{% for (key,value) in ansible_facts.packages.iteritems() %}
{{ value }}
{% endfor %}
But this ended I an error to:
fatal: [somehost.example.org]: FAILED! => {"changed": false, "msg": "AnsibleUndefinedVariable: 'list object' has no attribute 'name'"}
It seems I am not smart enough to figure out the solution, can someone lend me a hand?
Sincerely,
a frustrated Ansible user

Answer 1

In the data sample you posted, package name is the key and value is a list of 1 element comprising of a dictionary.
To get the version you have to access the first element. Like so:
{% for key, value in ansible_facts.packages.iteritems() %}
{{ key }} {{ value[0].version }}
{% endfor %}
Should render the file with package list like:
...
zlib 1.2.11
...