Support Lodash filters in the CLI? about temme HOT 2 CLOSED

I think using lodash functions as filter would be great. However, there is one major problem -- temme can only parse simple JavaScript literals while "lodash filters" requires complex arguments. For example, sortBy(collection, iteratees) requires iteratees a function, which is not a simple literal value and cannot be parsed by temme.

If you want process the output of temme, you can pipe the output from temme to jq. For example...

cat a.html | temme some-selector-or-selector-file | jq xxxx

Or save the temme-selector in a file and use that file in the CLI, so that you can define some customized filters to process data. For example...

echo 'div { $foo; }' > a.temme
temme a.temme "<div class=foo>hello</div>"
# output:  {"foo":"hello"}

from temme.

temme can only parse simple JavaScript literals while "lodash filters" requires complex arguments.

Ah, fair enough. That makes sense.

Or save the temme-selector in a file and use that file in the CLI, so that you can define some customized filters to process data.

Thanks. I'll do that.

One thing that might be worth considering is some kind of import or require statement that can pull in a JavaScript module which can interact with the library to add custom filters e.g.:

// movies.temme
import "filters.mjs";

// movies.temme
require("filters.js");

- and/or a command-line option to do this (like -r/--require in node/ruby etc.). The module could export a function which is passed the temme exports e.g.:

// filters.js

module.exports = function (temme) {
    temme.defineFilter(...)
}

Or possibly:

// filters.mjs

export default function ({ defineFilter }) { ... }

This could then be invoked with e.g.:

$ temme -r filters.js -s <selector-expression> -i <html-expression>
$ temme -r filters.mjs -S <selector-file> -I <html-file>

As you can see, the options would probably need to be made more explicit to accommodate this, but I'd actually prefer that anyway.

Closing this, thanks (but I'm happy to raise a new issue for the require/--require proposal — or discuss it here — if there's any interest).

By the way, thanks for temme! It's the best DSL for extracting data from unstructured documents I think I've ever used, and I've tried (and even written :-) quite a few over the years 👍

from temme.