aboutsummaryrefslogtreecommitdiff
path: root/doc/todo/toc-with-human-readable-anchors.mdwn
blob: 1e41add96c3ff9696439397fee8b7547ca804899 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
The [[/plugins/toc]] plugin is very useful but it creates anchors with names such as #index1h3

In #ikiwiki today, another user and I were in agreement that an option for human readable anchors would be preferable.

> +1 - i would love to see that happen too. Here's a patch I wrote a while back for similar functionality in moinmoin: https://svn.koumbit.net/koumbit/trunk/patches/moinmoin/nice_headings.patch -- [[anarcat]]

----

I started looking into this again after getting annoyed at the
unreadable anchors, and here's what I came up with.

[[!template  id=gitbranch branch=anarcat/toc-recycle-id author="[[anarcat]]"]]

The first step is to fix [[plugins/toc]] to use headings: we can
figure out how to generate those later, but it would be nice if the
toc directive would just reuse existing headings instead of relying on
its own. I do this by simply checking if there's a `id` field (which
is, by standard, unique) and reuse that when building the table of
contents. This requires parsing HTML element attributes, but that
shouldn't impact performance too much, hopefully. The old IDs are
still generated for backwards compatibility. This is done in
my [toc-recycle-id branch][] (see [921a264][]).

[921a264]: https://gitlab.com/anarcat/ikiwiki/commit/27d5d9d126b6b675ad273ebd63095df0c921a264
[toc-recycle-id branch]: https://gitlab.com/anarcat/ikiwiki/commits/toc-id-recycle

The second step is to generate those headings. There are two ways of
doing this:

 1. enable multimarkdown. by default, the [[plugins/mdwn]] plugin will
    add `id` anchors when using [Text::Multimarkdown][] which is
    simply a matter of adding `multimarkdown: 1` in the setup file

    > I don't think multimarkdown is a good solution. It served a useful
    > purpose when we were defaulting to [[!cpan Text::Markdown]] or to
    > `markdown.pl`, but now that we're using Discount by default,
    > Multimarkdown is mostly a trap for the unwary - it's a less predictable
    > and (in general) less featureful parser than Discount. Ideally we'd
    > always be using CommonMark or Discount these days, but as
    > far as I know there's still no API-stable CommonMark library. --[[smcv]]

    > > Sure - but then does discount introduce those identifiers in headings?
    > > And what about the patch to recycle those identifiers? --[[anarcat]]

 2. enable the [[plugins/headinganchors]] plugin. if multimarkdown is
    disabled, this can also provide usable identifiers.

An issue I had with the latter plugin was that it did not work if
multimarkdown was enabled, as it doesn't match headings if they
already have a `id` attribute. It also doesn't deal very well with
non-ASCII characters: they get basically garbled into their numeric
representation. I have therefore written a derivative of the
headinganchor plugin called [[plugins/contrib/i18nheadinganchors]] to
work around those issues.

It would be great to see the `toc` part of this patchset merged, at
least. It could also be a configurable option, but that seems overkill
considering that backwards compatibility is kept... --[[anarcat]]

[Text::Multimarkdown]: http://search.cpan.org/search?mode=dist&query=Text%3A%3AMarkdown

[[!tag wishlist patch]]