Handle ampersands

[jeremyhansonfinger]

Ampersands in the ruleset are currently ignored.

[wooorm]

@jeremyhansonfinger What’s the issue with ampersands? Maybe I can help?!

P.S. I’m recently adding some more plugins to retext. Might be something useful in there (for example, indefinite-article, repeated-words).

[jeremyhansonfinger]

@wooorm This is probably pretty simple, but I can't figure out how to properly escape special characters when creating a rule and writing a test for it.

So far I've tried &, \&, and %26.

Rule:

  "&": {
    "replace": "and",
    "note": "Don't use ampersands."
  },

Test:

test('ampersand', function (t) {
    var actual;

    t.plan(2);

    actual = process('bacon & Eggs.');
    t.equal(
        actual.messages[0].ruleId,
        '&',
        '“&” violates ruleId “&”'
    );
    t.equal(
        actual.messages[0].source,
        'retext-shopify',
        'source is “retext-shopify”'
    );
});

[jeremyhansonfinger]

It fails with TypeError: Cannot read property 'ruleId' of undefined

[wooorm]

Ah, alright. So, the deal is that nlcst-search searches for words only, and & isn’t a word (it’s a symbol).

First: are you trying to ban all ampersands? Is AT&T wrong too? Or are loose ampersands wrong (&), and/or use in words like so: &c.?

AT&T and &c. are both words, so you should be able to search for those.

Loose ampersands, are as mentioned before, symbols, and those must be searched for using different techniques.

Something like:

var visit = require('unist-util-visit');
var toString = require('nlcst-to-string');
// `file` and `tree` are given to the transformer.
visit(tree, 'SymbolNode', function (node) {
  if (toString(node) === '&') {
    file.warn('Do not use ampersands', node, 'ampersand');
  }
});

[jeremyhansonfinger]

Right, of course. I forgot about how the tree is constructed of punctuation elements and word elements.

[wooorm]

Yeah that's something to keep in mind!
Note that it's also possible to regex over the whole file, by coercing it to string ('file.toString'),
And checking that. Then, use vfile-location to get positions from offsets.