Pinned 2026 toolchains (Go 1.26, Rust 1.98/edition 2024, Python 3.14 + uv, Node 24, Zig 0.16, NixOS 26.05), postgres 18 / mongo 8, lockfiles built from, non-root runtimes, .dockerignore, per-project LICENSE, READMEs with the git.devai.io clone line, checkout@v7 CI. Security fixes in the legacy Rust APIs (any-password login, self-assigned admin, hard-coded JWT secret), JWT alg/exp/sub enforcement across the blog series, safe markdown links in the frontends, and many smaller bugs — every project was built, run and exercised end to end. Adds scripts/publish.sh + a CI publish job that splits every folder into its own repo at git.devai.io/templates/<folder>. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01128fhuZbgivaSJvtMf4s1G
65 lines
2.2 KiB
Markdown
65 lines
2.2 KiB
Markdown
# dictionary-lookup
|
|
|
|
Type a word and read its pronunciation and definitions, grouped by part of
|
|
speech, from a free dictionary API. It's your lesson in digging data out of
|
|
real JSON — arrays tucked inside objects tucked inside arrays.
|
|
|
|
## Run
|
|
|
|
Get it: `git clone https://git.devai.io/templates/dictionary-lookup.git`
|
|
|
|
Double-click `index.html` — it opens in your browser and works, as long as
|
|
you're online (it calls the dictionary live). Nothing to install.
|
|
|
|
Or serve it like production:
|
|
|
|
```sh
|
|
docker compose up --build
|
|
```
|
|
|
|
Then open http://localhost:8080.
|
|
|
|
## How it works
|
|
|
|
`lookup(word)` fetches
|
|
`https://freedictionaryapi.com/api/v1/entries/en/<word>` — English Wiktionary
|
|
as JSON, no key needed. The answer is nested three levels deep:
|
|
|
|
```
|
|
data.entries[] one per part of speech (noun, verb, …)
|
|
.senses[] one per meaning
|
|
.definition the text
|
|
.examples[] optional example sentences
|
|
```
|
|
|
|
`render()` walks down with one loop per level: an `<h3>` for each entry's
|
|
`partOfSpeech`, an `<ol>` of its senses, and the first example under a
|
|
definition when there is one. Everything goes onto the page with
|
|
`textContent`, never `innerHTML`, so text from the API can't inject HTML.
|
|
|
|
The states are handled too: "Loading…" while waiting, a friendly message when
|
|
the word isn't found (the API answers with an empty `entries` array), when
|
|
you hit the 1,000-lookups-an-hour limit (HTTP 429), or when the network fails.
|
|
|
|
Try it: show each sense's `synonyms`, cap each part of speech at two
|
|
definitions with `item.senses.slice(0, 2)`, or add a "Surprise me" button.
|
|
|
|
## Layout
|
|
|
|
```
|
|
index.html the search form, a status line and an empty entry card
|
|
app.js lookup() and render()
|
|
styles.css the look; follows your system's light or dark mode
|
|
```
|
|
|
|
## Deploy
|
|
|
|
Push to your own GitHub repo and the shipped workflow
|
|
(`.github/workflows/ci.yml`) tests the compose stack, publishes the image to
|
|
GHCR, and — once you set the `DEPLOY_HOST` / `DEPLOY_USER` variables and
|
|
`DEPLOY_KEY` secret — deploys it to your server over ssh.
|
|
|
|
---
|
|
Part of [devai.io](https://devai.io) — the APIs & Data track: fetch real data
|
|
from the internet. Next up:
|
|
[country-explorer](https://git.devai.io/templates/country-explorer).
|