Files
images/README.md
T

106 lines
3.5 KiB
Markdown

# Image Library
A single-page, **backend-free** landing page that lists every image in `img/`,
lets you full-text search by filename, and copies an image's URL to your
clipboard when you click it. Built to be dropped onto an nginx server as plain
static files.
```
index.html the page
assets/styles.css styling (auto light/dark)
assets/app.js search, rendering, clipboard logic
generate-manifest.py scans img/ -> images.json, then commits
pushAll regenerate + commit + push, in one command
images.json generated manifest the page fetches
img/ put your images here
```
## Why a manifest?
A browser cannot list a directory on its own, and "no backend" rules out an API
that would do it. So the file list is produced by a tiny build step:
`generate-manifest.py` walks `img/` and writes `images.json`. The page fetches
that file at runtime. Everything the server does is serve static files.
## Quick start
```sh
# 1. Add images (subfolders are fine)
cp ~/pictures/*.png img/
# 2. Build the manifest (and commit it)
./generate-manifest.py
# 3. Preview locally
python3 -m http.server 8080
# -> open http://localhost:8080
```
Re-run step 2 whenever you add, rename, or delete images.
## Publishing with `pushAll`
`pushAll` does the whole publish flow in one command: regenerate the manifest,
commit the changes, and push to the remote.
```sh
./pushAll
```
To run it as `pushAll` from anywhere, either add this folder to your `PATH`, or
create a git alias once:
```sh
git config alias.pushAll '!./pushAll'
# then:
git pushAll
```
### `generate-manifest.py` options
```sh
./generate-manifest.py # generate + commit
./generate-manifest.py --no-commit # generate only
./generate-manifest.py -m "Add logos" # custom commit message
```
## Features
- **Search** — case-insensitive, multi-word with AND semantics. Every word must
appear somewhere in the filename; prefix and word-boundary matches rank first.
Matches are highlighted. Press `/` to focus the search box, `Esc` to clear.
- **Click to copy** — clicking a card copies the image's absolute URL
(e.g. `https://your-host/img/logo.png`). A toast confirms it.
- **Lazy loading** — thumbnails load as you scroll, so large libraries stay fast.
- **No dependencies** — vanilla JS/CSS, no framework, no build tooling.
## Deploying on nginx
Copy the whole project (including the generated `images.json`) into a web root
and point a server block at it. See [`nginx.conf.example`](nginx.conf.example).
```sh
# build the manifest locally, then upload everything
./generate-manifest.py --no-commit
rsync -av --delete ./ user@server:/var/www/image-library/
```
If you'd rather generate the manifest on the server, install Python 3 there and
run the script after uploading images. Python is only needed for that step —
nginx never runs it.
### Clipboard over plain HTTP
`navigator.clipboard` only works in a **secure context** (HTTPS, or
`localhost`). The app detects this and falls back to a legacy copy method, which
works on plain HTTP in most browsers. For guaranteed behavior on a LAN host,
serve over HTTPS (e.g. a self-signed cert or a reverse proxy) — or just use
`http://localhost` while testing.
## Alternative: nginx autoindex
If you don't want a build step at all, you can enable nginx's directory listing
and have the page parse it, but that requires `autoindex on`, produces fragile
HTML scraping, and loses the manifest metadata. The manifest approach is
recommended.