AAP-30426: Lightspeed API playbook generation and explanation to accept a new custom_prompt field

[manstis]

Jira Issue: https://issues.redhat.com/browse/AAP-30426

Description

Support customPrompt JSON field for both Playbook Generation and Explanation.

Playbook Generation

• The backend expects {goal} or both {goal} and {outline} keywords to be present in the custom_prompt
• Depending on if Goal or both Goal and Outline are provided

Playbook Explanation

• The backend expects {playbook} keyword to be present in the custom_prompt

Testing

Steps to test

1. Pull down the PR
2. make start-backends
3. make create-application
4. Run server
5. Create User API token:

• Login into Lightspeed as somebody
• Login to http://localhost:8000/admin as Django admin
• Create an Access Token for the User you logged into Lightspeed as

6. Run generation tests:

• Valid "custom prompt"

curl -X 'POST' \
'http://localhost:8000/api/v0/ai/generations/' \
-H 'accept: application/json' \
-H 'Content-Type: application/json' \
-H 'Authorization: Bearer <access_token_created_above>' \
-d '{
"text": "Install apache",
"customPrompt": "You are an Ansible Expert. Generate a playbook to do {goal}"
}'

Response should be OK.

curl -X 'POST' \
'http://localhost:8000/api/v0/ai/generations/' \
-H 'accept: application/json' \
-H 'Content-Type: application/json' \
-H 'Authorization: Bearer <access_token_created_above>' \
-d '{
"text": "Install apache",
"outline": "1. Install apache",
"customPrompt": "You are an Ansible Expert. Generate a playbook to do {goal} along the lines of {outline}"
}'

Response should be OK.

• Invalid "custom prompt"

curl -X 'POST' \
'http://localhost:8000/api/v0/ai/generations/' \
-H 'accept: application/json' \
-H 'Content-Type: application/json' \
-H 'Authorization: Bearer <access_token_created_above>' \
-d '{
"text": "Install apache",
"customPrompt": "You are an Ansible Expert. Generate a playbook to do something"
}'

Response should represent:

{"customPrompt":["'{goal}' placeholder expected."]}

curl -X 'POST' \
'http://localhost:8000/api/v0/ai/generations/' \
-H 'accept: application/json' \
-H 'Content-Type: application/json' \
-H 'Authorization: Bearer <access_token_created_above>' \
-d '{
"text": "Install apache",
"outline": "1. Install apache",
"customPrompt": "You are an Ansible Expert. Generate a playbook to do {goal} along the lines of something"
}'

Response should represent:

{"customPrompt":["'{outline}' placeholder expected when 'outline' provided."]}

7. Run explanation tests:

• Valid "custom prompt"

curl -X 'POST' \
'http://localhost:8000/api/v0/ai/explanations/' \
-H 'accept: application/json' \
-H 'Content-Type: application/json' \
-H 'Authorization: Bearer <access_token_created_above>' \
-d '{
"content": "---\n- name: Install apache\n  hosts: all\n  tasks:\n    - name: Install apache\n      ansible.builtin.package:\n        name: httpd\n        state: present\n    - name: Start apache\n      ansible.builtin.service:\n        name: httpd\n        state: started\n",
"customPrompt": "You are an Ansible Expert. Explain playbook {playbook}"
}'

Response should be OK.

• Invalid "custom prompt"

curl -X 'POST' \
'http://localhost:8000/api/v0/ai/explanations/' \
-H 'accept: application/json' \
-H 'Content-Type: application/json' \
-H 'Authorization: Bearer <access_token_created_above>' \
-d '{
"content": "---\n- name: Install apache\n  hosts: all\n  tasks:\n    - name: Install apache\n      ansible.builtin.package:\n        name: httpd\n        state: present\n    - name: Start apache\n      ansible.builtin.service:\n        name: httpd\n        state: started\n",
"customPrompt": "You are an Ansible Expert. Explain playbook something"
}'

Response should represent:

{"customPrompt":["'{playbook}' placeholder expected."]}

Scenarios tested

As above.

Production deployment

• This code change is ready for production on its own
• This code change requires the following considerations before going to production:

See ansible/vscode-ansible#1545

[manstis]

I believe @TamiTakamiya may be changing the MODEL_ID/SPACE_ID used for Playbook Generation/Explanation.

That however does not affect the changes made here.

[TamiTakamiya]

I believe @TamiTakamiya may be #1324 the MODEL_ID/SPACE_ID used for Playbook Generation/Explanation.

My change is specifically for fixing AAP-30690, i.e. not to send the granite-3b model_id to WCA Playbook Explanation/Generation APIs. It should not break this.

[sonarqubecloud]

Quality Gate passed

Issues
1 New issue
0 Accepted issues

Measures
0 Security Hotspots
94.7% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarCloud

[TamiTakamiya]

LGTM! Thanks!