36863 - CSS: small-caps/uppercase: ?ß should become SS

Status: VERIFIED FIXED

(Core :: Internationalization, defect, P3)

Description

[Tobias Burnus]

Hi, this is with the M16 snapshoot from 2000-04-21-13.

font-variant:small-caps: the `ß' ß should IMO converted to `SS'
when used with small-caps or when converted to uppercase
(text-transform:uppercase).
This is at least convention in German (and as far as I know, the SS lignature
(wrongly called SZ lignature, but who cares) is only used in German).

Tobias
This war previously included in Bug #36827

Comment 1

[David Hallowell]

Confirming...

Comment 2

[Pierre Saslawsky]

Reassigned to ftang who is the last one to have touched csCRT::ToUpper().

Comment 3

[Frank Tang]

This require special casing handling. 

Comment 4

[Frank Tang]

related to 16796
Here is my change to nsICaseConversion.h so we now have a new set of method 
which can handle the case the input string have different length of the output 
string.

I need pierre/troy's help to integrate into nsTextTransformer and nsTextFrame

I will check in the following code soon.

Index: unicharutil//public/nsICaseConversion.h
===================================================================
RCS file: /m/pub/mozilla/intl/unicharutil/public/nsICaseConversion.h,v
retrieving revision 1.4
diff -u -r1.4 nsICaseConversion.h
--- nsICaseConversion.h 1999/11/06 03:26:12     1.4
+++ nsICaseConversion.h 2000/04/25 21:31:41
@@ -25,6 +25,7 @@
 
 #include "nsISupports.h"
 #include "nscore.h"
+#include "nsString.h"
 
 // {07D3D8E0-9614-11d2-B3AD-00805F8A6670}
 #define NS_ICASECONVERSION_IID \
@@ -55,6 +56,11 @@
   // Convert an array of Unicode characters into title case
   NS_IMETHOD ToTitle( const PRUnichar* anArray, PRUnichar* aReturn, 
                       PRUint32 aLen, PRBool aStartInWordBundary=PR_TRUE) = 0;
+
+  // The following nsString flavor one know how to handle special casing
+  NS_IMETHOD ToUpper(const PRUnichar* anIn, PRUint32 aLen, nsString& anOut, 
const PRUnichar* aLocale=nsnull) = 0;
+  NS_IMETHOD ToLower(const PRUnichar* anIn, PRUint32 aLen, nsString& anOut, 
const PRUnichar* aLocale=nsnull ) = 0;
+  NS_IMETHOD ToTitle(const PRUnichar* anIn, PRUint32 aLen, nsString& anOut, 
const PRUnichar* aLocale=nsnull, PRBool aStartInWordBoundary=PR_TRUE) = 0;
    
 };
 
Index: unicharutil//src/nsCaseConversionImp2.cpp
===================================================================
RCS file: /m/pub/mozilla/intl/unicharutil/src/nsCaseConversionImp2.cpp,v
retrieving revision 1.13
diff -u -r1.13 nsCaseConversionImp2.cpp
--- nsCaseConversionImp2.cpp    2000/04/05 00:10:08     1.13
+++ nsCaseConversionImp2.cpp    2000/04/25 21:31:41
@@ -333,6 +333,96 @@
   return NS_OK;
 }
 
+#define k_ss         0x00df
+#define kDot_I       0x0130
+#define kDot_i       PRUnichar('i')
+#define kDotLess_I   PRUnichar('I')
+#define kDotLess_i   0x0131
+
+NS_IMETHODIMP nsCaseConversionImp2::ToUpper
+  (const PRUnichar* anIn, PRUint32 aLen, nsString& anOut, const PRUnichar* 
aLocale) 
+{
+  anOut.SetString(anIn,aLen);
+
+  // Special casing - Turkish dotless I
+  if((nsnull != aLocale ) && (PRUnichar('t')==aLocale[0]) && (PRUnichar('r') == 
aLocale[1]))
+  {
+     for(PRUnichar* s=(PRUnichar*)anOut.GetUnicode(); *s ; s++)
+     {
+        if(kDot_i == *s)
+             *s = kDot_I;
+     }
+  }
+
+  ToUpper(anOut.GetUnicode(), (PRUnichar*)anOut.GetUnicode(), anOut.Length());
+
+  // Special casing - SS
+  PRInt32 idx=0;
+  for(PRUnichar* s=(PRUnichar*)anOut.GetUnicode(); *s ; s++,idx++)
+  {
+     if(k_ss == *s) {
+        *s = PRUnichar('S') ; 
+        anOut.Insert(PRUnichar('S'),idx);
+        // Insert may cause reallocate, so we need to GetUnicode() again
+        s = (PRUnichar*)anOut.GetUnicode() + idx;
+        idx++;
+     }
+  }
+  return NS_OK;
+}
+NS_IMETHODIMP nsCaseConversionImp2::ToLower
+  (const PRUnichar* anIn, PRUint32 aLen, nsString& anOut, const PRUnichar* 
aLocale) 
+{
+  anOut.SetString(anIn,aLen);
+
+  // Special casing - Turkish dotless I
+  if((nsnull != aLocale ) && (PRUnichar('t')==aLocale[0]) && (PRUnichar('r') == 
aLocale[1]))
+  {
+     for(PRUnichar* s=(PRUnichar*)anOut.GetUnicode(); *s ; s++)
+     {
+        if(kDot_I == *s)
+             *s = kDot_I;
+     }
+  }
+
+  ToLower(anOut.GetUnicode(), (PRUnichar*)anOut.GetUnicode(), anOut.Length());
+
+  return NS_OK;
+}
+NS_IMETHODIMP nsCaseConversionImp2::ToTitle
+  (const PRUnichar* anIn, PRUint32 aLen, nsString& anOut, const PRUnichar* 
aLocale,
+              PRBool aStartInWordBoundary)
+{
+  anOut.SetString(anIn,aLen);
+
+  // Special casing - Turkish dotless I
+  if((nsnull != aLocale ) && (PRUnichar('t')==aLocale[0]) && (PRUnichar('r') == 
aLocale[1]))
+  {
+     for(PRUnichar* s=(PRUnichar*)anOut.GetUnicode(); *s ; s++)
+     {
+        if(kDot_i == *s)
+             *s = kDot_I;
+     }
+  }
+
+  ToTitle(anOut.GetUnicode(), (PRUnichar*)anOut.GetUnicode(), anOut.Length(),
+              aStartInWordBoundary);
+
+  // Special casing - SS
+  PRInt32 idx=0;
+  for(PRUnichar* s=(PRUnichar*)anOut.GetUnicode(); *s ; s++,idx++)
+  {
+     if(k_ss == *s) {
+        *s = PRUnichar('S') ; 
+        anOut.Insert(PRUnichar('S'),idx);
+        // Insert may cause reallocate, so we need to GetUnicode() again
+        s = (PRUnichar*)anOut.GetUnicode() + idx;
+        idx++;
+     }
+  }
+  return NS_OK;
+}
+
 
 
 nsCaseConversionImp2::nsCaseConversionImp2()
Index: unicharutil//src/nsCaseConversionImp2.h
===================================================================
RCS file: /m/pub/mozilla/intl/unicharutil/src/nsCaseConversionImp2.h,v
retrieving revision 1.7
diff -u -r1.7 nsCaseConversionImp2.h
--- nsCaseConversionImp2.h      1999/11/06 03:26:15     1.7
+++ nsCaseConversionImp2.h      2000/04/25 21:31:41
@@ -51,6 +51,9 @@
   NS_IMETHOD ToTitle(const PRUnichar* anArray, PRUnichar* aReturn, 
                      PRUint32 aLen, PRBool aStartInWordBoundary = PR_TRUE);
    
+  NS_IMETHOD ToUpper(const PRUnichar* anIn, PRUint32 aLen, nsString& anOut, 
const PRUnichar* aLocale=nsnull) ;
+  NS_IMETHOD ToLower(const PRUnichar* anIn, PRUint32 aLen, nsString& anOut, 
const PRUnichar* aLocale=nsnull );
+  NS_IMETHOD ToTitle(const PRUnichar* anIn, PRUint32 aLen, nsString& anOut, 
const PRUnichar* aLocale=nsnull, PRBool aStartInWordBoundary=PR_TRUE) ;
 private:
   static nsrefcnt gInit;
 };

Comment 5

[Frank Tang]

troy, I will come to your office and discuss w/ you about the current 
nsTextTransformer code. I am not sure what is the best way to integrate the new 
method call.

Comment 6

[Frank Tang]

changing the case coversion code in the intl/unicharutil won't help. Currently, 
the code in nsTextFrame.cpp call the character based (not string based) ToUpper 
from nsCRT. I try to fix the nsTextFrame.cpp but it is very hard because we need 
to expand the buffer and also the position need to align with the space buffer. 
(search ToUpper in nsTextFrame.cpp to locate the code, there are two, the 
measure width one is easier to fix. but the draw string one is very hard without 
hurting the performance.)

Comment 7

[Frank Tang]

add [e-mail] to cc list.

Comment 8

[Frank Tang]

add test cases into test0.html but hide in comment.
The uppercase part is fixed in bug 16796. limit the scope of this bug to the 
small-caps for easier bug tracking.

Comment 9

[Frank Tang]

case conversion issue. Reassign to nhotta and cc erik.
It is easier to fix this problem in the nsTextFrame.cpp code to deal with this 
special case. I have fix it for text-transform: uppercase in 
nsTextTransformer.cpp
This is really layout issue.

Comment 10

[Frank Tang]

reassign to erik

Comment 11

[Frank Tang]

want to fix this?

Comment 12

[Teruko Kobayashi]

Changed QA contact to [e-mail] for now.

Comment 13

[Shanjian Li]

Attachment: proposed patch

Comment 14

[Shanjian Li]

erik, need your sr=

Comment 15

[Erik van der Poel]

Shanjian, I see some code for CSS's "small-caps", but what about
"text-transform: uppercase"?

Comment 16

[Shanjian Li]

Frank mentioned thad transform:uppercase has been fixed.

Comment 17

[Erik van der Poel]

r=erik

Comment 18

[Shanjian Li]

fix checked in.

Comment 19

[Andreas Becker]

verified fixed using the following builds:
Win98 2001032204
Mac 2001032608
Linux 2001032608