Migration von WordPress zu Hugo: Markdown mit OpenAI-API

import { existsSync } from "https://deno.land/std/fs/mod.ts";
import OpenAI from 'npm:openai@latest';

function* walkDirectory(start, path) { // das ist eine nützliche rekursive Generator-Funktion :-)
    for (const entry of Deno.readDirSync(`${start}/${path ?? ''}`)) {
        const entryPath = `${path && path.length > 0 ? `${path}/` : ''}${entry.name}`;
        if (entry.isFile) {
            yield entryPath;
        }
        if (entry.isDirectory) {
            yield* walkDirectory(start, entryPath);
        }
    }
}

const openai = new OpenAI({
    apiKey: Deno.env.get('OPENAI_API_KEY'),
});

async function convertToMarkdown(html) {
    const chatCompletion = await openai.chat.completions.create({
        messages: [
            {
                role: 'system',
                content: 'You are provided with HTML code. Your task is to transform it to its closest Markdown representation. Before conversion, remove any HTML comments in the input. An image "test.jpg" should be represented as "![A test image](test.jpg)" where "A test image" is the alt or title attribute and "test.jpg" is the src attribute.',
            },
            {
                role: 'user',
                content: html,
            }
        ],
        model: 'gpt-3.5-turbo',
    });
    console.log({ chatCompletion, });
    return chatCompletion.choices[0].message.content;
};

const htmlPath = 'out/html';
const markdownPath = 'out/md';

for (const filePath of walkDirectory(htmlPath)) {
    if (!filePath.endsWith('.html')) {
        continue;
    }

    const filename = `${htmlPath}/${filePath}`;
    const newFilename = `${markdownPath}/${filePath.replace(/\.html/, '.md')}`;

    if (existsSync(newFilename)) {
        continue;
    }

    const contents = await Deno.readTextFile(filename);

    console.log({
        input: filename,
        contents,
    });

    const newContents = await convertToMarkdown(contents);

    console.log(newContents);

    await Deno.writeTextFile(newFilename, newContents);
}

OPENAI_API_KEY="" deno run --allow-read=./out --allow-write=./out/md --allow-env=OPENAI_API_KEY --allow-net=api.openai.com convert-to-markdown-with-openai.js

<h1>Lorem Ipsum</h1>

<div id="navigation">
    <ul>
        <li><a href="#abschnitt1">Abschnitt 1</a></li>
        <li><a href="#abschnitt2">Abschnitt 2</a>
            <ul>
                <li><a href="#abschnitt2a">Teil A</a></li>
                <li><a href="#abschnitt2b">Teil B</a></li>
            </ul>
        </li>
        <li><a href="#abschnitt3">Abschnitt 3</a></li>
    </ul>
</div>

<div id="content">
    <div id="abschnitt1">
        <h2>Abschnitt 1</h2>
        <p><strong>Lorem</strong> ipsum dolor sit amet.</p>
        <img src="bild1.jpg" alt="Bild 1">
        <ul>
            <li>Punkt 1</li>
            <li>Punkt 2
                <ul>
                    <li>Unterpunkt 2a</li>
                    <li>Unterpunkt 2b</li>
                </ul>
            </li>
        </ul>
    </div>

    <div id="abschnitt2">
        <h2>Abschnitt 2</h2>
        <div id="abschnitt2a">
            <h3>Teil A</h3>
            <p><strong>Sed</strong> cursus ante dapibus diam.</p>
            <img src="bild2.jpg" alt="Bild 2">
        </div>
        <div id="abschnitt2b">
            <h3>Teil B</h3>
            <table>
                <tr>
                    <th>Spalte 1</th>
                    <th>Spalte 2</th>
                </tr>
                <tr>
                    <td><img src="bild3.jpg" alt="Bild 3"></td>
                    <td><strong>Daten 2</strong></td>
                </tr>
            </table>
        </div>
    </div>

    <div id="abschnitt3">
        <h2>Abschnitt 3</h2>
        <p>Nullam quis <strong>risus</strong> eget urna mollis ornare.</p>
        <ol>
            <li>Erster Schritt</li>
            <li>Zweiter Schritt
                <img src="bild4.jpg" alt="Bild 4">
            </li>
        </ol>
    </div>
</div>

# Lorem Ipsum

## Abschnitt 1
**Lorem** ipsum dolor sit amet.
![Bild 1](bild1.jpg)
- Punkt 1
- Punkt 2
  - Unterpunkt 2a
  - Unterpunkt 2b

## Abschnitt 2

### Teil A
**Sed** cursus ante dapibus diam.
![Bild 2](bild2.jpg)

### Teil B

| Spalte 1 | Spalte 2 |
|---------|---------|
| ![Bild 3](bild3.jpg) | **Daten 2** |

## Abschnitt 3
Nullam quis **risus** eget urna mollis ornare.
1. Erster Schritt
2. Zweiter Schritt ![Bild 4](bild4.jpg)