wpconvert/convert.py

import yaml
import click
import os
import dotenv
import mistletoe
import re
import requests
import tempfile
import mimetypes
import base64

from xmlrpc.client import Binary
from urllib.parse import urlparse

image_cache = {}

def upload_image(image_path):

    global image_cache

    if image_path in image_cache:
        return image_cache[image_path]
    
    site_url = os.environ['WP_ENDPOINT']
    username = os.environ['WP_USER']
    password = os.environ['WP_PASSWORD']

    # Initialize the WordPress XML-RPC client
    wp = Client(site_url, username, password)

    mime = mimetypes.guess_type(image_path)[0]
    basename = os.path.basename(image_path)

    print("url", image_path)
    print("basename", basename)
    print("mimetype", mimetypes.guess_type(image_path)[0])

    with open(image_path,'rb') as f:
        thumb =  wp.call(media.UploadFile({
            "name": basename,
            "type": mime,
            "bits": Binary(f.read())
        }))

    image_cache[image_path] = thumb

    return image_cache[image_path]

def reupload_image(image_url):

    global image_cache

    if image_url in image_cache:
        return image_cache[image_url]

    site_url = os.environ['WP_ENDPOINT']
    username = os.environ['WP_USER']
    password = os.environ['WP_PASSWORD']

    # Initialize the WordPress XML-RPC client
    wp = Client(site_url, username, password)

    r = requests.get(image_url)
    url = urlparse(image_url)
    basename = os.path.basename(url.path)

    mime = mimetypes.guess_type(image_url)[0]

    print("url", url)
    print("basename", basename)
    print("mimetype", mimetypes.guess_type(image_url)[0])

    thumb =  wp.call(media.UploadFile({
        "name": basename,
        "type": mime,
        "bits": Binary(r.content)
    }))

    with open(basename,'wb') as f:
        print("write content")
        f.write(r.content)

    image_cache[image_url] = thumb

    return image_cache[image_url]

def replace_figure_tags(markdown_file, html_string):

    html_string = html_string.replace("{{<figure", "<figure")
    html_string = html_string.replace(">}}",">")

    # Define a regular expression pattern to match the original <figure> tag with optional alt and caption
    pattern = r'<figure src="([^"]+)"(?: alt="([^"]+)")?(?: caption="([^"]+)")?>'

    # Define a function to create the replacement string
    def replace_figure(match):
        src, alt, caption = match.groups()
        replacement = '<figure>\n'
        if src:

            if src.startswith('http'):
                new_url = reupload_image(src)['url']
            else:
                relpath = os.path.join(os.path.dirname(markdown_file), src)
                path = os.path.abspath(relpath)
                new_url = upload_image(path)['url']

            replacement += f'  <img src="{new_url}"'
        if alt:
            replacement += f' alt="{alt}"'
        replacement += '>\n'
        if caption:
            replacement += f'<figcaption>{caption}</figcaption>'
        replacement += '</figure>'
        return replacement

    # Use re.sub to find and replace all <figure> tags in the input string
    new_html_string = re.sub(pattern, replace_figure, html_string)

    return new_html_string



from wordpress_xmlrpc import Client, WordPressPost
from wordpress_xmlrpc.methods import posts, media



@click.command()
@click.argument('markdown_files', nargs=-1, type=click.Path(file_okay=True, dir_okay=False))
def upload_hugo_posts_to_wordpress(markdown_files):

    dotenv.load_dotenv()

    # Configure your WordPress site URL and credentials
    site_url = os.environ['WP_ENDPOINT']
    username = os.environ['WP_USER']
    password = os.environ['WP_PASSWORD']

    # Initialize the WordPress XML-RPC client
    wp = Client(site_url, username, password)

    # Change to the directory where your Hugo Markdown posts are located
    #import os
    #os.chdir(post_directory)

    # Get a list of Markdown files in the directory

    # for markdown_file in markdown_files:
    #     # Read the Hugo Markdown file
    for markdown_file in markdown_files:
        with open(markdown_file, 'r', encoding='utf-8') as file:
            content = file.read()

        # separate header from body
        _,frontmatter, body = content.split("---\n", maxsplit=2)

        fm = yaml.safe_load(frontmatter)

        print(fm)


        # Parse your Markdown file to extract title and content (you may need to use a library like frontmatter or Hugo-specific logic here)

        body = replace_figure_tags(markdown_file, body)

        # Create a WordPress post
        post = WordPressPost()
        post.slug = fm['url']
        post.title = fm['title']
        post.date = fm['date']
        post.content = mistletoe.markdown(body)


        if 'thumbnail' in fm:
            if fm['thumbnail'].startswith('http'):
                post.thumbnail = reupload_image(fm['thumbnail'])['id']
            else:
                relpath = os.path.join(os.path.dirname(markdown_file), fm['thumbnail'])
                path = os.path.abspath(relpath)
                post.thumbnail = upload_image(path)['id']

        print(post.content)
        # Add additional fields like categories, tags, etc. as needed

        # Publish the post to WordPress
        wp.call(posts.NewPost(post))

if __name__ == '__main__':
    upload_hugo_posts_to_wordpress()