• Published:

My blog moved from WordPress to Hugo back in 2023. Hugo had a clear appeal: it was cheap. I could host the site on GitHub and avoid paying for a VPS at all. But it also had a drawback that became more annoying over time — I couldn't just sit down and write wherever I happened to be.

At first I ran Hugo on the computer at my shop. That meant writing only after I got to the shop, or drafting somewhere else and then uploading from the shop machine. Either way, the process was broken into pieces. Writing often depends on momentum, and once that flow is interrupted, the desire to finish a post tends to fade.

Later I moved the Hugo setup onto my VPS. Now I could write from anywhere, but there was still friction: I had to open a terminal, connect to the VPS, and go through too many steps before I could actually write. I lost count of how many times I thought about going back to WordPress. The trouble was that every tutorial online seemed to go in the opposite direction — WordPress to Hugo. The people writing those guides seemed to have no interest in returning to WordPress.

Then I came across Ghost. I installed it with Docker and tried it out, and it felt good. It wasn't as heavy as WordPress, and it had an admin panel where I could write posts directly. I decided to move the blog to Ghost. Searching for a migration guide didn't help, so a fellow blogger suggested I ask AI for help.

I had taught myself a little Python during the pandemic, so I decided to convert the Hugo Markdown files into a JSON file that Ghost could import. With ChatGPT's help, I finally got the code working:

import os
import json
import re
import yaml
import markdown
from datetime import datetime

def parse_markdown(filepath):
    with open(filepath, 'r', encoding='utf-8') as f:
        text = f.read()

    match = re.match(r'^---\n(.*?)\n---\n(.*)', text, re.DOTALL)
    if not match:
        print(f"⚠️ 跳过文件：{filepath}，没有找到 front matter")
        return None

    front_matter_text, content_part = match.groups()

    try:
        front_matter = yaml.safe_load(front_matter_text)
    except Exception as e:
        print(f"⚠️ YAML 解析错误：{filepath} -> {e}")
        return None

    title = front_matter.get('title')
    raw_date = front_matter.get('date')
    tags = front_matter.get('tags', [])

    if not title or not raw_date or not content_part.strip():
        print(f"⚠️ 数据不完整，跳过文件：{filepath} -> title: {title}, date: {raw_date}, content_length: {len(content_part.strip())}")
        return None

    try:
        if isinstance(raw_date, datetime):
            dt = raw_date
        elif isinstance(raw_date, str):
            try:
                dt = datetime.fromisoformat(raw_date.replace("Z", "+00:00"))
            except ValueError:
                try:
                    dt = datetime.strptime(raw_date, "%Y-%m-%d %H:%M:%S")
                except ValueError:
                    dt = datetime.strptime(raw_date, "%Y-%m-%dT%H:%M:%S%z")
        else:
            print(f"⚠️ 日期格式无法识别，跳过：{filepath} -> date: {raw_date}")
            return None
    except Exception as e:
        print(f"⚠️ 日期解析失败：{filepath} -> {e}")
        return None

    date_str = dt.strftime('%Y-%m-%dT%H:%M:%S.000Z')
    html_content = markdown.markdown(content_part.strip(), extensions=['extra', 'codehilite', 'tables'])

    return {
        'title': title,
        'slug': os.path.splitext(os.path.basename(filepath))[0],
        'html': html_content,
        'created_at': date_str,
        'updated_at': date_str,
        'tags': tags
    }

def walk_md_files(base_dir):
    for root, _, files in os.walk(base_dir):
        for file in files:
            if file.endswith('.md'):
                yield os.path.join(root, file)

def main():
    base_dir = './'
    print(f"开始遍历目录：{base_dir}")
    posts = []
    tags_set = set()

    for filepath in walk_md_files(base_dir):
        print(f"📄 正在处理文件：{filepath}")
        post = parse_markdown(filepath)
        if post:
            post_tags = []
            for tag_name in post['tags']:
                tag_str = str(tag_name).strip()
                if tag_str:
                    tags_set.add(tag_str)
                    post_tags.append(tag_str)
            post['tags'] = post_tags
            posts.append(post)

    ghost_data = {
        "meta": {
            "exported_on": int(datetime.now().timestamp() * 1000),
            "version": "5.0.0"
        },
        "data": {
            "posts": [],
            "tags": [],
            "posts_tags": [],
            "users": [
                {
                    "id": 1,
                    "name": "admin",
                    "slug": "admin",
                    "email": "[email protected]"
                }
            ]
        }
    }

    for i, post in enumerate(posts, start=1):
        ghost_data["data"]["posts"].append({
            "id": i,
            "title": post["title"],
            "slug": post["slug"],
            "html": post["html"],
            "status": "published",
            "created_at": post["created_at"],
            "updated_at": post["updated_at"],
            "published_at": post["created_at"],
            "author_id": 1
        })

    tag_id_map = {tag: idx + 1 for idx, tag in enumerate(tags_set)}
    for tag, tag_id in tag_id_map.items():
        ghost_data["data"]["tags"].append({
            "id": tag_id,
            "name": tag,
            "slug": tag.lower().replace(" ", "-")
        })

    for post_id, post in enumerate(posts, start=1):
        for tag in post["tags"]:
            ghost_data["data"]["posts_tags"].append({
                "post_id": post_id,
                "tag_id": tag_id_map[tag]
            })

    with open("ghost_import.json", "w", encoding="utf-8") as f:
        json.dump(ghost_data, f, ensure_ascii=False, indent=2)

    print("✅ 已生成 ghost_import.json 文件")

if __name__ == "__main__":
    main()

To use it, save the script as Hugotoghost.py and put it in the content folder. Then run:

python hugotoghost.py

It will produce a ghost_import.json file. Upload that file in the Ghost admin panel and the posts should be imported.

Related Posts