[openbeos] more CharacterSet and StyledEdit news

- CharacterSet news

Thanks to Axel and Ingo for their advice on restructuring the CharacterSet and 
CharacterSetRoster.  I did as we discussed and created a new library 
libtextencoding.so.  It's built 
from the sources in src/kits/support.  It does have convert_to_utf8 and 
convert_from_utf8 
functions but they are currently simply stubs.  In addition to those, it has 
BCharacterSet and 
BCharacterSetRoster which are defined in the BPrivate namespace.

BCharacterSet is very similar to my earlier proposal, except that it now has a 
CountAliases() 
function and a AliasAt(uint32 index) function.  It also has two new 
constructors, a default 
constructor, and a constructor providing all the fields.  The class is doxygen 
documented in the 
header, which is at private/support/CharacterSet.h.

BCharacterSetRoster is a bit different from before.  I followed Ingo's advice 
and made it a 
lightweight stateful object. (one int)  It performs iterator through the 
functions 
GetNextCharacterSet(BCharacterSet * charset) and RewindCharacterSets().  I 
rewrote 
StyledEdit to use these iterators.

BCharacterSetRoster also has 3 static functions for retrieving character sets 
in constant time. 
(from FontID, ConversionID, or MIBenum)  It has two functions for retrieving 
character sets in 
linear time by PrintName or by standard name. (IANA, MIME, or a known alias)  
The retrieval 
functions have not been well exercised.  

Lastly BCharacterSetRoster has two possibly superfluous functions: 
StartWatching and 
StopWatching.  I put these in because of my concern that we are going to have 
an unweildy 
number of character sets.  Presenting all of these character sets in menus in 
Terminal, StyledEdit, 
BeMail, etc., could get ugly.  It's not too much of a problem yet but I was 
thinking perhaps it could 
be possible for us to have a character set pref that would allow the user to 
select the character 
sets that they commonly use.  Of course, such a pref would be live and if the 
user changed that set, 
then applications which have encodings menus would like to be notified so they 
can change 
their menus appropriately.  Since this would obviously be an R2 feature I have 
left it with stub 
implementations (return B_ERROR).  Also the whole idea needs to be discussed, 
etc.

- StyledEdit news

And for those watching, I did promise some more StyledEdit news and here it is. 
 Printing works 
now.  I mean, works, like, works real good.  You can even print selected pages. 
 Print in color, 
print wacky fonts, etc.  The behavior is dpi perfect to R5 StyledEdit as far as 
I can tell, even 
down to the buggy printing of right-aligned text.  (Perhaps OBOS BTextView 
isn't buggy this 
way?)

In addition to printing I added in cancelling via escape on the find/replace 
windows. (thanks 
Technix for pointing that out)  I've also revamped the find window so that it 
is font sensitive.  
The window doesn't change its size in response to the font size (maybe it 
should?)  But the items 
on it are all layed out so that they will be legible at sizes up to about 18.  
(For those wanting to set 
their system font to 13+ pixels I point you to FontSelector on bebits: 
http://www.bebits.com/app/
267 )  Compare the behavior with R5 StyledEdit and you can appreciate it 
better. :-)

The replace window hasn't been made font-sensitive yet though.  Also, the 
border effect on the 
OBOS Find window is slightly different than the R5 Find window.  I can't figure 
out how to 
replicate the R5 effect yet.  All my pixel tweaks have failed so far.

As I said before StyledEdit has been changed to use the new 
CharacterSet-related headers and 
implementation.  Since there isn't a working convert_to_utf8, etc., yet for 
libtextencoding.so, I 
linked the relevant files directly into StyledEdit.  So, OBOS StyledEdit is 
very usable now on R5.  
It's also a little bit bigger than it needs to be, but still smaller than the 
R5 StyledEdit. :-)  As usual, 
I'm sharing a built copy on beshare:StyledEdit.zip@Andrew and it's all in cvs 
of course.

Andrew Bachmann

Other related posts: