androb / polyglotproxy

AI Tools for TinyMCE

This repo is a collection of experiments with the OpenAI API.

polyglotproxy's Issues

Create Insert Illustration feature

Break up the image generation feature into illustration, photo, and art. This will enable more refined prompts and improved usability, as many business users are likely to want an illustration(?)

Mindmap creation

Create a nested list of a mindmap to insert into the document

Enable different models

Provide a setting to switch between the different models from OpenAI, HuggingFace, Replicate, and Anthropic.

Implement Annotations for Writing Suggestions

Use TinyMCE's Annotations API to mark up writing suggestions from OpenAI. This may only work on GPT-4.

Below, I've provided the marked-up paragraph with the writing suggestions as comments using TinyMCE's annotation format:

Herald of Free Enterprise is often referred to as one of the <span data-mce-comment="1">worse</span><ins>worst</ins> tragedies of <span data-mce-comment="2">time..</span><ins>time.</ins> From the <span data-mce-comment="3">face</span><ins>surface</ins> of it all, the tragedy appears to have been a case of <span data-mce-comment="4">negligency</span><ins>negligence</ins> by officials working on the vessel. However<ins>,</ins> the cardinal faults <span data-mce-comment="5">laid</span><ins>lay</ins> higher up in the company<ins>:</ins> social issues such as <span data-mce-comment="6">ignorance towars</span><ins>ignorance towards</ins> adopting a safety <span data-mce-comment="7">cultur</span><ins>culture</ins>, not adhering to standards and <span data-mce-comment="8">profeedures</span><ins>procedures</ins>, and unclear definition of roles were all faults contributing to the death of 193 passengers.

To annotate the text with TinyMCE, you would need to register a custom annotation type and then apply the annotations using the annotate method. Here's an example of how you might do this:

// Register a custom annotation type for writing suggestions
tinymce.activeEditor.annotator.register('writing-suggestion', {
  decorate: function(uid, data) {
    return {
      attributes: {
        'data-mce-writing-suggestion': uid
      },
      classes: ['my-writing-suggestion']
    };
  }
});

// Apply the writing suggestions as annotations
tinymce.activeEditor.annotator.annotate('writing-suggestion', { suggestionId: 1, content: 'change to correct comparative form' });
tinymce.activeEditor.annotator.annotate('writing-suggestion', { suggestionId: 2, content: 'fix punctuation' });
tinymce.activeEditor.annotator.annotate('writing-suggestion', { suggestionId: 3, content: 'improve word choice' });
tinymce.activeEditor.annotator.annotate('writing-suggestion', { suggestionId: 4, content: 'spelling correction' });
tinymce.activeEditor.annotator.annotate('writing-suggestion', { suggestionId: 5, content: 'change to correct verb form' });
tinymce.activeEditor.annotator.annotate('writing-suggestion', { suggestionId: 6, content: 'spelling correction' });
tinymce.activeEditor.annotator.annotate('writing-suggestion', { suggestionId: 7, content: 'spelling correction' });
tinymce.activeEditor.annotator.annotate('writing-suggestion', { suggestionId: 8, content: 'spelling correction' });

This example registers a custom annotation type called 'writing-suggestion' and applies it to the text using the annotate method with the appropriate suggestion IDs and content. Note that the example assumes the TinyMCE editor is already initialized and active.