[Ohrrpgce] Text markup, again

Sat Jun 6 02:27:21 PDT 2020

bbcode was my preference too. I've been thinking about how backcompat could
be maintained with a bit, and I've decided it's just too much trouble.
Instead I propose the syntax [.col3]...[/col]. That's just as easy to type
as bbcode, because even international keyboard layouts will all have easy
access to ".", and very unintrusive (even more so if using a proportional
font.) Or as an alternative, [:col3]...[/col]. I propose [/col] instead of
[./col] because there's no need for the extra character. More details at
the end.

I wrote an rpgbatch script to scan the gamelists for strings. I only
scanned textboxes, items, attacks, global text strings, and slices. Some
totals (probably a lot of duplicate games):

num textboxes: 658473 attacks: 101674 items: 336286
num scripts with strings: 4967
strings containing [ or ]: 22539

And scanning for valid tags (eg no internal spaces) like [b], [/b], <b>,
</b>, [.b], <.b> I found:
[foo] tags: 3722
<foo> tags: 482
[.foo] tags: 0
<.foo> tags: 0
[/foo] tags: 2
</foo> tags: 2
games containing [...] or <...>: 225

Of the two uses of [/...],  one was "[/Demo]" in Detelamane.rpg and the
other was in my "1 day in IRC" game!
   [01:27] spoonweaver> [IMG]
http://i46.photobucket.com/albums/f117/spoonweaver/Tim-Tim0016.png[/IMG]
Of the two uses of </...>, "</whisper>" and again 1day irc.rpg:
  [11:25] C9X[Homework]> <blink> <img src="facepalm.jpg"> </blink>

Now, <blink> is actually a markup I might add some day, but 1day irc.rpg
used its own text system using box border sprites (those strings are in
scripts), so those occurrences are irrelevant.

I also scanned for some other character sequences

Many uses of [$ especially as e.g. [${C0}]
4 games used \]
4 games used [! (never as would-be-valid markup like [!foo])
Two games used <! (never as would-be-valid markup)
One game used <. (never as would-be-valid markup)
12 games used [. particularly for "[...]" and e.g. "[...no?]"  (never as
would-be-valid markup)
No games used [:
A handful of games had binary characters < 32 due to corrupt data. Also
some contained ASCII 13 (\r), apparently we didn't strip those out when
importing textboxes.

So a more complete specification:

Markup looks like [.<TAG><ARGS>] or [/<TAG>]
TAG: A tag name is case insensitive, containing only a-zA-Z.
Tags would usually have a short and a long spelling, eg sp and space, b and
bold.
ARGS: Some tags take one or more arguments.
An argument is either a number, or a string not containing ], comma or
whitespace.
ARGS can optionally start with =, mandatory for string arguments
Multiple arguments are separated with commas.

If a tag doesn't meet the above syntax or isn't a recognised tag name or
has the wrong number of arguments then it's displayed literally.

[/foo] doesn't need to follow [foo]. It undones the last [foo], and tags
can be closed in a different order than they were opened.
If a tag can't be undone (e.g. [space]) then the closing tag is ignored and
displayed literally.

Escaping: [\. can be used to display as [.
\[. would be a bad choice of escape sequence because appending markup to
arbitrary text wouldn't be safe.
I probably wouldn't bother to actually implement this escape code until
later.

On Tue, 2 Jun 2020 at 10:31, James Paige <Bob at hamsterrepublic.com> wrote:

> I have been thinking about this today.
>
> There are pros and cons for different markup styles, but
> [color=10][/color] with backcompat-optional bbcode markup would probably be
> most familiar to people, and easiest to type. [img#.#] for images/sprites.
> [sp#] for horizontal spacing
>
> Any of the styles will actually be fine-- but if you need help breaking
> the tie and settling on one, that is my 2 cents :)
>
> ---
> James
>
>
>
> On Mon, Jun 1, 2020 at 5:55 AM Ralph Versteegen <teeemcee at gmail.com>
> wrote:
>
>> I want to finally expose the ability to use text markup and non-8x8 fonts
>> to users. Even before all the menus and slice collections are updated, and
>> before the textbox file format or any others are replaced. Therefore there
>> will be graphical problems and shortcomings, but we need that in order to
>> find and fix what needs fixing. It can be marked experimental.
>>
>> (I have a plan for textboxes: just repurpose the 304 bytes used to store
>> the existing 8 lines of text as a single autowrapped 304 byte string, plus
>> an extra 200 bytes or so tacked onto the end of each SAY record so there's
>> plenty of spare space to actually use markup/compact fonts.)
>>
>> But this means we need to finalise the markup. I'm very undecided. Sorry,
>> lots of waffle.
>>
>> Previously:
>>
>> On Friday, January 27, 2017, Ralph Versteegen <teeemcee at gmail.com> wrote:
>>
>>>
>>> On 28 January 2017 at 13:10, Ralph Versteegen <teeemcee at gmail.com>
>>> wrote:
>>>
>>>> Also, I decided that it's best not to use ${...} codes for markup
>>>> supported by printstr such as color and font. ${...} should be exclusively
>>>> for inserting a string variable in the text, as in many programming
>>>> languages. Instead I was thinking of using #{...}, or maybe quite different
>>>> syntax which has (optional) closing tags, like $[col45]...$[/col], since
>>>> closing tags really are needed (right now I'm using -1 to mean 'restore
>>>> previous'). Other tag names could include colbg, pal, spc (no closing tag),
>>>> font, indent, indentr, and even link=.
>>>>
>>>
>>> Plus also no-argument $[b], $[i], $[h] tags, which switch to the
>>> bold/italic/highlight font.
>>> Actually, maybe we can just use [b], [col45], etc, or <b>, <col45>...
>>> and add a backcompat bit to enable them, and use a $ prefix to escape them
>>>
>>
>> Godot also uses bbcode:
>> https://docs.godotengine.org/en/stable/tutorials/gui/bbcode_in_richtextlabel.html#reference
>>
>> I like the idea of allowing both short and long forms of markup, eg both
>> [c10] and [col10] and [color10], because there will be a lot of codes. But
>> it makes parsing harder. So I'll just list long forms below.
>>
>> Options include:
>> -#{color10} #{/color}
>> -or some other variant like :{color10}, !{color10}.
>> -[color=10] [/color]  or [color10]    (= is optional)
>> -just [color=10] [/color]  (= is mandatory if there's an arg)
>> -$[color10] $[/color]
>> -<color10> </color>, or <color=10> </color>. Not quite XML.
>> -RPG Maker uses \C[10] and a lot of other one-character backslash codes
>> (see attachment) for other things like pauses.
>>
>> Two-argument tags are also required, for embedding image n frame m. E.g.
>> <spr4,8>
>>
>> I already notice that I find it really easy to accidentally type ${i}
>> instead of #{i}.
>>
>> Markup beginning with ! or with : has a problem: the ! or : is easily
>> mistaken as part of the preceding text, eg "Gold:{C10} ${V2}"
>>
>> $ is used for string operators and embed codes, and I wanted the syntax
>> $"Gold: ${gold}" to embed a local or global variable by name. so Using $
>> neerd
>>
>> Note that the text renderer doesn't process markup that doesn't have a
>> valid tag name or argument. So there's less backcompat risk
>>
>> Nonetheless If we were to use bbcode, a backcompat bit to disable it
>> seems warranted. If we used anything else, like #{i} then I see no need for
>> a bit. That would simplify a lot of code since we wouldn't need to pass the
>> backcompat bit as "withtags" all over the codebase, which seems pretty
>> horrible, and especially for text slices in slice collections. Should
>> probably rule out bbcode just for this, unless the backcompat only affected
>> textboxes (and maybe plotstrings) and nothing else. That's probably already
>> good enough to avoid all backcompat problems. I guess I should run
>> rpgbatch.py to extract textboxes and other strings and see what I find.
>>
>> bbcode or <color10> are easiest to type. That's important.
>>
>> bbcode or angle brackets may be a nuiscance on forums, wikis, or help
>> pages.
>>
>> And then I also want to support mediawiki markup, for help pages. That
>> would be an optional per-slice setting, available to users too. But
>> mediawiki isn't a popular markup. Markdown is. Honestly if easy typing is a
>> concern, optional markdown seems like the ultimate solution for
>> italics/bold/underline/strikethrough/indent. This comparison of markup
>> languages is interesting:
>> http://hyperpolyglot.org/lightweight-markup#text-style
>>
>> And then there's the tag names to use.
>> [i] should mean italic, but I also wanted i for icon. Hmm, ico or icon.
>> And img for sprites... or spr? spr is probably too similar to spc, for
>> horizontal space (in pixels). I suppose s would be strike-through not space
>> or sprite?
>>
>>
>> _______________________________________________
>> Ohrrpgce mailing list
>> ohrrpgce at lists.motherhamster.org
>> http://lists.motherhamster.org/listinfo.cgi/ohrrpgce-motherhamster.org
>>
> _______________________________________________
> Ohrrpgce mailing list
> ohrrpgce at lists.motherhamster.org
> http://lists.motherhamster.org/listinfo.cgi/ohrrpgce-motherhamster.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.motherhamster.org/pipermail/ohrrpgce-motherhamster.org/attachments/20200606/f76f2d84/attachment.html>