--- title: "Consolidating 18 Years of Writing: From 6 Platforms to One" description: "From Blogspot to Drupal to WordPress to Octopress to 11ty and finally Astro - the story of consolidating 372 posts across 6." url: "https://h3manth.com/scribe/consolidating-6-platforms-18-years/" canonical_url: "https://h3manth.com/scribe/consolidating-6-platforms-18-years/" markdown_url: "https://h3manth.com/scribe/blog/consolidating-6-platforms-18-years.md" publishDate: "2025-12-27T00:00:00.000Z" author: "Hemanth HM" category: "web" tags: ["javascript","python","astro"] word_count: 959 reading_time_minutes: 5 x_markdown_tokens: 1671 content_signal: "ai-train=yes, search=yes, ai-input=yes" --- # Consolidating 18 Years of Writing: From 6 Platforms to One > From Blogspot to Drupal to WordPress to Octopress to 11ty and finally Astro - the story of consolidating 372 posts across 6. **TL;DR:** Consolidated 18 years of blog content from 6 different platforms (Blogspot → Drupal → WordPress → Octopress → 11ty → Astro) into a single static site with 372 posts, recovered 200+ posts from database dumps and backups, removed 36 duplicates, and fixed 220+ posts with broken formatting - all with AI pair programming. ## The Blogging Timeline ``` 2007 ──► Blogspot (Google's free platform, where it all began) │ 2008 ──► Drupal 6 (self-hosted, felt like a "real" developer) │ 2012 ──► WordPress (everyone was doing it) │ 2014 ──► Octopress (Ruby-based static site generator, hipster phase) │ 2020 ──► 11ty (JavaScript ecosystem, modern SSG) │ 2025 ──► Astro (the final form? ) ``` Each migration carried forward content from the previous platform. Some posts survived all 6 migrations. Most accumulated formatting quirks along the way. ## Why So Many Migrations? Each platform served its purpose at the time: - **Blogspot** - Zero config, just write and publish. Perfect for a student. - **Drupal** - Wanted more control, PHP was everywhere. - **WordPress** - Plugins for everything! (Until you need to patch 47 of them) - **Octopress** - "Static sites are the future!" Also, I was learning Ruby. - **11ty** - JavaScript everywhere, simple and fast. - **Astro** - Type-safe content collections, island architecture, and actual good DX. ## The Great Consolidation The content was scattered across: - A Drupal 6 SQL dump (~50MB) - WordPress exports (XML) - Octopress markdown files - 11ty markdown files The Drupal content was the messiest - 18 years of different editors, plugins, and format changes had left scars in the data. ### The Drupal Challenge Parsing the SQL dump required some Python wizardry: ```python def parse_drupal_nodes(sql_file): with open(sql_file, 'r') as f: content = f.read() # Parse INSERT statements for node_revisions pattern = r"INSERT INTO `node_revisions` VALUES\s*\((.*?)\);" matches = re.findall(pattern, content, re.DOTALL) return parse_values(matches) ``` The URL aliases needed 322 redirect rules: ```apache RewriteRule ^content/(.*)$ /scribe/$1 [R=301,L] RewriteRule ^node/(.*)$ /scribe/$1 [R=301,L] ``` ### The Code Block Nightmare Each platform handled code blocks differently: - Drupal used `` tags with custom filter modules - WordPress had its own syntax highlighters - Octopress used `{% codeblock %}` liquid tags - 11ty used standard fenced blocks The result after consolidation? **Fragment fences everywhere:** ````markdown ```javascript function foo() { ```javascript ← WHY return bar; ```javascript ← STOP } ``` ```` ### The Fix Scripts Two Python scripts saved the day: **Language detection:** ```python def detect_language(code): if re.search(r'^\s*import\s+\w+', code, re.MULTILINE): return 'python' if re.search(r'\.each\s+do\s*\|', code): return 'ruby' if re.search(r'^\s*\$\s+\w+', code, re.MULTILINE): return 'bash' return 'javascript' ``` **Fragment fence removal:** ```python patterns = [ (r'(\n[^\n`]+)\n```javascript\n([^\n`]+\n)', r'\1\n\2'), (r'(\n[^\n`]+)\n```python\n([^\n`]+\n)', r'\1\n\2'), ] ``` ## The Statistics | Metric | Value | |--------|-------| | Platforms migrated from | 6 | | Total posts consolidated | 372 | | Posts recovered from DB/backup | 200+ | | URL redirects (100% working) | 292 | | Posts cleaned (nav/tweet links) | 220 | | Duplicate posts removed | 36 | | Years of content | 18 | | Build time | ~6 seconds | | Final static site size | ~91MB | ## The AI Pair Programming Stats The entire migration was completed over **multiple sessions** with AI pair programming. Here's the breakdown: | Metric | Value | |--------|-------| | Total project time | ~4-5 hours (across sessions) | | AI tool calls | 3800+ | | Posts analyzed | 400+ | | Posts recovered/fixed | 250+ | | Python scripts written | 10+ | | Deployments | 20+ | | Git commits | 30+ | ### The AI Workflow 1. **Human reports broken post** → AI views file 2. **AI rewrites with proper code blocks** → Human tests locally 3. **Human approves** → AI commits and deploys 4. **Repeat** for 70+ posts ### Time Saved Without AI, manually fixing 120+ posts with code block issues would have taken: - **~5-10 min per post** to view, analyze, and rewrite - **120 posts × 7 min = ~14 hours** of manual work **AI turned a 2-day project into a few hours of evening sessions.** ## What I've Learned 1. **Content outlives platforms** - Posts from 2007 are still getting traffic. The platform is temporary; the content is permanent. 2. **Each migration leaves scars** - Every platform has its quirks. Migrating content 5+ times means 5+ layers of formatting weirdness. 3. **Static sites win** - No more security patches. No more database maintenance. Just HTML served from edge locations. 4. **AI pair programming is a game-changer** - What would have been a weekend project became an evening session. Writing migration scripts, fixing broken posts, iterating on solutions - AI made it possible. 5. **Beware of auto-fix scripts** - A script that "fixes" code blocks across 200+ files can break more than it fixes. Some posts needed manual restoration after aggressive auto-fixing. Always test on a subset first. 6. **Live databases are gold** - The SQL dump wasn't enough. Connecting to the live MySQL database recovered 13 more posts that were missing from the static dump. 7. **llms.txt matters** - Added [h3manth.com/llms.txt](https://h3manth.com/llms.txt) for AI crawlers. If bots are reading our content anyway, might as well make it structured. ## The Final Architecture ``` h3manth.com ├── /scribe/ → 372 blog posts (Astro) ├── /ai/ → AI projects ├── /fun/ → Interactive demos ├── llms.txt → AI-friendly site description └── *.html → Landing pages ``` The blog is now live at [h3manth.com/scribe](https://h3manth.com/scribe) with all 372 posts, proper syntax highlighting, and 292 redirects preserving 18 years of SEO juice. **From 6 disparate platforms to one unified Astro site - it's been an evolution. Here's to the next 18 years.** --- *P.S. - If you're migrating legacy content, don't underestimate the code block problem. And get yourself an AI pair programmer.*