Problems with string.gsub

[gumphrey]

I am trying to get plain text out of chat messages. I want to strip all the color codes, link information, etc. just leaving a plain text version of the message. Is there an existing method exposed that can do that?

I have been experimenting with using string.gsub to remove the text, but it doesn't work except for very simple patterns. For instance, in stand alone lua I can use this pattern to find item links and replace them with just plain text in square brackets. In my addon, it does nothing...

local textOut = string.gsub(textIn, 'H.-h(%[.-%])h', '%1')

[gumphrey]

After some more experimentation, I think the problem is the % codes. Is something different used? I tried "normal" regex syntax, and that doesn't seem to work either...

[gumphrey]

Nevermind, % codes work sometimes. It just seems busted when you do certain things (different than stock Lua which makes no sense to me), and it is rather unpredictable.

[Sharlikran]

function ZO_LinkHandler_ParseLink(link)
    if type(link) == "string" then
        local color, data, text = link:match("|H(.-):(.-)|h(.-)|h")
        return text, color, zo_strsplit(':', data)
    end
end

That is for a totally different thing, but I was wondering if it might help. It might strip out everything. See if it works, if not I am not good at Lua yet so I can't offer advice. That code was to me from someone else.

[Garkin]

Originally Posted by gumphrey I am trying to get plain text out of chat messages. I want to strip all the color codes, link information, etc. just leaving a plain text version of the message. Is there an existing method exposed that can do that? I have been experimenting with using string.gsub to remove the text, but it doesn't work except for very simple patterns. For instance, in stand alone lua I can use this pattern to find item links and replace them with just plain text in square brackets. In my addon, it does nothing... local textOut = string.gsub(textIn, 'H.-h(%[.-%])h', '%1')

Item link starts with "|H" then it contains clolor code, link type etc and ends with "|h" then there is item name (or just text to disply) and "|h" again. If I'm not mistaken, there are no [] in item name, so you should change pattern.

local textOut = string.gsub(textIn, "|H.-|h(.-)|h", "%1")

I'm not in game, so I did not try it.

[ins]

There are two types of itemlinks.
With or without [].

* GetItemLink(*integer* _bagId_, *integer* _slotIndex_, *[LinkStyle|#LinkStyle]* _linkStyle_)

[Sharlikran]

Originally Posted by ins There are two types of itemlinks. With or without []. * GetItemLink(*integer* _bagId_, *integer* _slotIndex_, *[LinkStyle|#LinkStyle]* _linkStyle_)

YES! I have been wondering about that. I will have to revisit this. I have been really bothered that sometimes when I parse the link I get "corn flower" and other times I get "corn flower^p" I bet it's that link style.

[Iyanga]

Originally Posted by Sharlikran YES! I have been wondering about that. I will have to revisit this. I have been really bothered that sometimes when I parse the link I get "corn flower" and other times I get "corn flower^p" I bet it's that link style.

These meta-characters are part of nearly all strings, even crafting station names have them and they are not an item nor can they be linked

[gumphrey]

Thanks everyone, that clears up a lot. The part I missed was the vertical bar, it was tricky to debug since putting any string out in any way to display ends up interpreting the vertical bars. Indeed everything works now and Lua is not behaving inconsistently as I first thought. Is there a raw dump which doesn't interpret all the strings (I may write a helper function for that now that I know what is going on)

Also out of curiosity what are all the other numbers in the link?

The links I see seem to be of the form:
|H<color>:<type?>:<id?>:<n1>:..:<n19>|h<linkname>|h

The color is clear. What are the possibilites for type (I am assuming it is the type of thing getting linked... I usually see "item" there). Is the next number the ID for that type (item ID for instance)? Anyone know what all the other numbers are (I generally see 19 more, but maybe that is dependent on what is getting linked?)

[ins]

See my reply:
http://www.esoui.com/forums/showpost...3&postcount=13

or:
http://wiki.esoui.com/ZO_LinkHandler_CreateLink

For information on item links.

[gumphrey]

Perfect, thanks!

[Garkin]

Originally Posted by Iyanga These meta-characters are part of nearly all strings, even crafting station names have them and they are not an item nor can they be linked

You can get rid of that characters using zo_strformat function, see this info:
http://www.esoui.com/forums/showthread.php?p=5007

[Seerah]

To clarify on why those characters are there, it's so that all the text around that item name (or npc name) can match.

^m, ^f, ^n = masculine, feminine, neutral (so surrounding text can switch to his, her, etc. or German can use the appropriate der, das, die, etc.)
^p = plural (so surrounding text can switch between it, them, etc.)

Are there others I'm forgetting?

[Iyanga]

Originally Posted by Seerah To clarify on why those characters are there, it's so that all the text around that item name (or npc name) can match. ^m, ^f, ^n = masculine, feminine, neutral (so surrounding text can switch to his, her, etc. or German can use the appropriate der, das, die, etc.) ^p = plural (so surrounding text can switch between it, them, etc.) Are there others I'm forgetting?

There are multi-letter meta-characters like ^ms. So it might be masculine singular, but then it's strange that not all meta-characters use two characters.