ji Reporter
	Description • 25 years ago

There is one option allowing users to convert the emoticons to graphics. 
But if the emoticons directly follow JPN chars without space, those emoticons
won't be converted to graphics. 

Steps of reproduce:
1. Lauch messenger.
2. Select Edit | Preferences.
3. Select Mail and Newsgroups | Messages
4. Check on Convert emoticons to graphics.
5. Compose a new message with the following lines in the msg body:
   日本語:)
   日本語 :)
6. Send the mail to the testing account itself.
   After receiving, you'll see the first smiley is not converted to 
   graphical happy face while the 2nd one is.

It's common for Asian users to use emoticons following text without space.

	nhottanscp
	Comment 1 • 25 years ago

cc to rhp

	Katsuhiko Momoi
	Comment 2 • 25 years ago

Please also debug with more than one ASCII spaces between
the end of a JPN string and the beginning of an emoticon
sequence, e.g. JPN_Sring     :)

I have seen some cases earlier where spaces made a difference
whether or not emoticons appear.

	Katsuhiko Momoi
	Comment 3 • 25 years ago

QA assingn --> ji

	nhottanscp
	Comment 4 • 25 years ago

Looks like this is not Japanese specific.
If I send, Hi:), it's not converted to emoticon either. Need spaces before :).
It this a spec?

Rich, please reassign to the developer of this feature, thanks.

	rhp (gone)
	Comment 5 • 25 years ago

Hi Ben,
Can you provide guidance on this one :-)

- rhp

	Ben Bucksch (:BenB)
	Comment 6 • 25 years ago

The specs can be found at <http://www.bucksch.org/1/projects/mozilla/16507>.



This is failure by intention :). If we don't check for delimitere, we'll mess up

e.g. >I:-) (a woman with an interesting hat) and be bashed by many other customers.



Does somebody have an idea, how to address both interests? Otherwise, I can do

nothing else than mark this WONTFIX.

	Ben Bucksch (:BenB)
	Comment 7 • 25 years ago

P5, because I'm not the maintainer of this class anymore.

	ji Reporter
	Comment 8 • 25 years ago

Changed the summary since it's not a Japanese specific problem.

	bobj
	Comment 9 • 25 years ago

I don't know how this is implemented, but it should properly parse word breaks
(simple white space parsing does not work for all languages), possibly use the
nsIWordBreaker class.

	Ben Bucksch (:BenB)
	Comment 10 • 25 years ago

> nsIWordBreaker class



Cool! If I were aware of this class, I would have used it from the beginning (I

wasn't even aware of the issue).



Are there other similar classes (other than nsLineBreak), which I should be

aware of, e.g. something like FindSentence?



BTW: nsIWordBreak needs documentation. I don't even know, what all the params

are for. (I know TestLineBreak, but still...)



I would /really/ like to fix this bug, but the new maintainer is Netscape (the

long story is, that a Netscape employee messed the class up so much, that I

*can't* maintain it anymore), and Netscape has no interest in enhancing the

class. However, I'll ACCEPT this bug for now and see, how things will develop.

	Frank Tang
	Comment 11 • 25 years ago

BreakInBetween
If we have two pieces of text ( "Th" + "is is a book")
BreakInBetween('Th", 2, "is is a book". 12, &out);
out will be false since it is NOT a word break between "Th" and "is is a book".
This is useful for text which logically continue but physically seperate . For example, "<B>Th</B>is is a book" is logically "This is a book" but physically belong to two text node "Th" and "is is a book"

In other hand, either BreakInBetween('This", 4, " is a book". 10,, &out); or BreakInBetween('This ", 5, "is a book". 9, &out); will be true

Next is used to find the "Next" word boundary
For example, Next("This is a book", 14, 0, &out,&needmore);
out will be 4 since the break position is in 4
needmore will be false because we can decide that the word break in position 4.
In the other hand,Next("This is a book", 14, 10, &out,&needmore);
needmore will be true since from the context, it cannot decide the "book" end after 'k' because it could be the "book" of the "bookcase" and need caller to pass in more context 

You should be able to handle them by just using these two methods. 

	Ben Bucksch (:BenB)
	Comment 12 • 25 years ago

ftang, tnx for clarification. Can somebody check this into the tree

(mozilla/intl/lwbrk/public/nsIWordBreaker.h), so others can take advantage of

it, too?

	Charles Wang
	Comment 13 • 25 years ago

IM and mail shows the same behavior on 03-24-17a jab1 build on Win98-Ja

	Ben Bucksch (:BenB)
	Comment 14 • 25 years ago

-> helpwanted :-(

	Serge Gautherie (:sgautherie)
	Comment 15 • 16 years ago

Filter on "Nobody_NScomTLD_20080620"

	Makoto Kato [:m_kato]
	Comment 16 • 15 years ago

still reproduce on Thunderbird 3.1