I built firsttripchina.com — a China travel guide written by someone who lives here

1436941541 — Wed, 29 Apr 2026 13:55:25 +0000

👋 Hey dev.to — I just shipped firsttripchina.com, a practical China travel guide written by someone who actually lives in China (me — Yunjie, a software engineer in Chengdu).

It's a content site for first-time foreign visitors, covering the friction points existing English-language China resources consistently get wrong: the 240-hour visa-free transit policy, how to register Alipay with a foreign card, the real-name attraction booking
system, and structured city itineraries (4-day Beijing, 3-day Zhangjiajie, Beijing → Xi'an by
HSR). No tour packages. No affiliate scams. Free to read, no signup.

## Why I built it

Every Western friend who's ever told me they were planning a China trip sent me the same five questions:

"Will my Visa card actually work?"
"Do I need a VPN?"
"Is the high-speed rail bookable in English?"
"How do I book Forbidden City tickets?"
"Do I need a tour, or can I do it solo?"

The English-language China travel internet is in a strange state. Lonely Planet's last meaningful update to its China chapter was years ago. TripAdvisor's China forum is full of well-intentioned travelers giving each other 2018 advice. The big content sites (Travel China Guide, China Highlights) are mostly OTA fronts pushing tour packages.

None of them know that the 240-hour visa-free policy expanded in late 2024, that Alipay now accepts foreign cards but with a ¥200 single-transaction limit and 3% fee, or that buying a Forbidden City ticket requires a real-name passport-linked booking 7 days in advance. I write Markdown for fun, I have a Vercel account, so I built the version I wished
existed.

## The tech stack (briefly)

Stack is React 19 + Vite + TypeScript on the frontend, Vercel Functions + Neon Postgres on the backend, deployed on Vercel.

The "obvious" content-site setup is content/posts/*.md parsed at build time — Astro or Hugo would have been the textbook answer. I went with Postgres rows for content because I want to edit from a hosted admin UI, not commit-and-deploy. The tradeoff is real (you can't git diff a content edit), but for a solo project the speed of "open admin, edit, save"
beats the audit trail.

For SEO I prerender the SPA to static HTML at build time — Puppeteer pulls every published slug from Postgres, visits each URL on a local vite preview server, and writes the rendered HTML into dist/<path>/index.html. Vercel then serves the static HTML before falling back to the SPA. This took longer than expected to get running on Vercel's build
container (you need puppeteer-core + @sparticuz/chromium, not regular Puppeteer, because the container doesn't ship the shared libraries Chromium expects). Each route runs through a usePageSeo hook that imperatively injects per-route <title>, meta tags, canonical, and JSON-LD, so prerender captures complete SEO HTML for every page.

The result: every page on the site lands in Google's index as static HTML — important, because Googlebot only executes JavaScript ~70% of the time on first crawl.

## What's currently live

12 destination pages (Beijing, Shanghai, Suzhou, Hangzhou, Zhangjiajie, Chengdu, plus more) and 8+ deep-dive blog posts.

Coming soon: Xi'an, Guilin, Yunnan destinations; a China high-speed rail master guide; a DiDi (rideshare) walkthrough for foreign passport holders.

The site is at firsttripchina.com — browse the destinations or blog and take a look.

I built a privacy-first PDF dark mode converter that runs entirely in your browser

1436941541 — Sat, 25 Apr 2026 08:58:01 +0000

TL;DR: I shipped pdfdark.org — a browser-side PDF dark mode converter. Files don't get

uploaded; the entire conversion happens in your browser via PDF.js + a Web Worker. Open source (MIT), free, no signup.

Why I built this

Reading long PDFs at night was killing my eyes. The two paths I had both sucked:

Existing "dark mode PDF" web tools → required uploading the file. For research papers, contracts, medical records

— that felt sketchy. No way to verify what they did with my data.

OS-level "invert colors" → wrecked photos and charts. Faces became X-rays. Graphs became noise.

So I built one with two non-negotiable defaults:

Nothing leaves your browser. PDF.js parses the file, a Web Worker does the dark-mode pass, pdf-lib stitches the

result. Verify it yourself in DevTools → Network.

Images keep their original colors. A saturation classifier detects photos and figures and leaves them untouched

while it darkens text and UI.

How it works

Drop PDF

→ PDF.js renders pages to canvas

→ Web Worker classifies each pixel by saturation

├─ saturated pixels (images, charts) → preserved

└─ low-saturation pixels (text, UI) → themed

→ pdf-lib assembles a new PDF

→ User downloads

The classifier runs on OffscreenCanvas inside a Web Worker, so the UI thread never blocks on large PDFs.

Output is a real PDF (image-based, one JPEG per page), so the dark mode persists when you email it, sync it to iPad, or

open it on a Kindle. Not a viewer toggle.

What it cost to ship

Domain: $7.50 (Cloudflare Registrar, .org)
Hosting: $0 (Vercel free tier)
Email forwarding: $0 (Cloudflare Email Routing)
Error monitoring: $0 (Sentry free tier, errors only with full content masking)
Time: a few weekends

Total cost: $7.50.

What I'm watching

Whether the "no-upload" angle resonates beyond privacy nerds
Edge cases that break the algorithm (weird embedded fonts, scanned PDFs)
Whether to add a vector-preserving mode for text-only PDFs (current output is image-based, so text isn't selectable)