Export for OpenCharacters

[drusepth]

Based on https://github.com/josephrocca/OpenCharacters/issues/35 it looks like we can export characters in a way that lets users talk to them with the following json form:

{
  name: "Character name",
  avatar: "avatar_url",
  personality: "In just a few words, how would your character describe themselves?",
  description: "In a few sentences, how would your character describe themselves? (Description of personality and other characteristics.)",
  scenario: "A few short sentences describing the starting scenario, circumstances, and context of the dialogue.",
  char_greeting: "The first message the character will say verbatim to start the conversation.",
  example_dialogue: "<START> [...] <START> [...]"
}

Following this character editor/exporter's format, we should probably also include a block of metadata in each export similar to:

    "metadata": {
        "version": 1,
        "created": 1681425505383,
        "modified": 1681425505383,
        "source": null,
        "tool": {
            "name": "AI Character Editor",
            "version": "0.5.0",
            "url": "https://zoltanai.github.io/character-editor/"
        }
    }

I'm gonna use this issue to plan out how to map a character's fields into these expected fields.

[drusepth]

For a given character @character:

Header	Header
name	@character.name
avatar	@character.random_image_including_private.src
personality	@character.name is a overview/gender overview/role, age overview/age (also known as overview/aliases).
description	@character.name is a social/occupation. Their background is history/background. Their primary motivations are nature/motivations. Their mannerisms are nature/mannerisms. Their flaws are nature/flaws and their prejudices are nature/prejudices. Their talents are nature/talents and their hobbies are nature/hobbies.
scenario	TBD (maybe just ask the user what they want the scenario to be?)
char_greeting	TBD
example_dialogue	TBD

Questions to ponder:

• Instead of building our own template from specific fields, should we just provide some kind of template and mapping for each field and let users pick and choose which fields they want to include in their character export?
  • Due to context window limitations (and the unpredictable variance in length on any particular field), it doesn't seem like including all data is feasible. Maybe we can pass in summaries of each category instead?

[drusepth]

Done & live.