IMG

IMG is a simple image stream generated using mk(1).

TODO:

HTML-escape alt text and captions.
Limit alt text and caption sizes.
Image file extensions should be recognized regardless of case.
The _timestamp files are used by feed.json and index.atom but not index.hml.
Note and link the Apple Shortcut I use to post from mobile.
index.html should link to the readme and (at least in the header) feed.json and index.atom.
Fix this page's css to indent ol items a bit.

Using it

One time-setup:

Satisfy the prerequisites.
Stick the mkfile in a directory your web server can get at.
Create files "_header" and "_footer" with whatever you like.
The header should start with something like "<html>" or whatnot, and must end with a line containing exactly and only the string "". The footer, similarly, must begin with a line containing exactly and only the string "", and should end the document with something like a "</html>". Mine are very simple and you can use them as examples for both header and footer, but you are free to use them to implement navigation, a sidebar, include javascript, whatever.
If you want to use the JSON or Atom feeds, you'll want _atomhead and _jsonhead. Change the strings in there to make sense for your installation.
Optionally, add an entry in cron to run "mk" in this directory.
In case you forget to run it yourself after making changes or creating a new entry. I run it every 10 minutes.

To create an entry:

Create an image file with a recognized extension.
We recognize the extensions jpeg, jpg, gif, and png.
Optionally, add alt text (highly encouraged) and/or a caption.
Create a test file with the same name as the image, plus ".alt" for the alt text and ".caption" for the caption.
Run 'mk' (optional if you created the 'cron' entry)..
This will add any new image file names to _timestamps if this is the first time we're seeing it, and then regenerate "index", "index.html", "index.atom", and "feed.json", as needed.

Other notes

Prerequisites:

On Plan 9:

You'll need my datefmt and statprograms.

On unix:

(This is expected to work but is completely untested.)
You'll need Plan 9 from User Space installed, and in your $path or $PATH before the normal unix tools.
You'll also need datefmt and stat, like above.

Reserved words/patterns:

Anything starting with an underscore ("_").
The files "index", "index.html", "index.atom", and "feed.json" will be updated (overwritten) by the mkfile.
Anything other file not ending with a recognized image file extension, .alt, or .caption should just be ignored.

Notes:

Unless your web server makes other arrangements, the full contents of this directory, including the mkfile, source, and names starting with _, will be available. That's a feature.
The JSON feed lacks context_text because I don't have a good way to safely escape it yet. I'm also not sure whether that should be alt text or captions, with or without fallback. Want to weigh in?
The _timestamps file is currently used inconsistently. The JSON and Atom feeds sorts by the entry in _timestamps, use that for 'date_published' or 'published', and use the file modification time for 'date_modified' or 'updated'. The index.html sorts by file modification time and doesn't expose the original published time.
plan9port support is entirely untested. I'd appreciate success/failure reports.

Inspired by the version at Texto-Plano.