| Subject: | 1.01 don't detect Latin2 Hungarian / Win1250 Hungarian / TIS620 Thai charset |
I write libchardet for C/C++ with Encode-Detect C++ binding source. And
It will find http://cvs.oops.org/?cvsroot=OOPS-
Project&module=libchardet.
On porting time, I found some bug of Encode-Detect 1.01 that can't
detect Latin2 Hungarian / Win1250 Hungarian / TIS620 Thai charset, And
these charset missing on SBCMGroupProber.
So, I report fixed patch.
Thanks.
| Subject: | 1.01.patch |
--- nsSBCSGroupProber.cpp 2009/02/21 20:08:02 1.1.1.1
+++ nsSBCSGroupProber.cpp 2009/02/23 13:47:40 1.2
@@ -56,21 +56,24 @@
mProbers[7] = new nsSingleByteCharSetProber(&Win1253Model);
mProbers[8] = new nsSingleByteCharSetProber(&Latin5BulgarianModel);
mProbers[9] = new nsSingleByteCharSetProber(&Win1251BulgarianModel);
+ mProbers[10] = new nsSingleByteCharSetProber(&Latin2HungarianModel);
+ mProbers[11] = new nsSingleByteCharSetProber(&Win1250HungarianModel);
+ mProbers[12] = new nsSingleByteCharSetProber(&TIS620ThaiModel);
nsHebrewProber *hebprober = new nsHebrewProber();
// Notice: Any change in these indexes - 10,11,12 must be reflected
// in the code below as well.
- mProbers[10] = hebprober;
- mProbers[11] = new nsSingleByteCharSetProber(&Win1255Model, PR_FALSE, hebprober); // Logical Hebrew
- mProbers[12] = new nsSingleByteCharSetProber(&Win1255Model, PR_TRUE, hebprober); // Visual Hebrew
+ mProbers[13] = hebprober;
+ mProbers[14] = new nsSingleByteCharSetProber(&Win1255Model, PR_FALSE, hebprober); // Logical Hebrew
+ mProbers[15] = new nsSingleByteCharSetProber(&Win1255Model, PR_TRUE, hebprober); // Visual Hebrew
// Tell the Hebrew prober about the logical and visual probers
- if (mProbers[10] && mProbers[11] && mProbers[12]) // all are not null
+ if (mProbers[13] && mProbers[14] && mProbers[15]) // all are not null
{
- hebprober->SetModelProbers(mProbers[11], mProbers[12]);
+ hebprober->SetModelProbers(mProbers[14], mProbers[15]);
}
else // One or more is null. avoid any Hebrew probing, null them all
{
- for (PRUint32 i = 10; i <= 12; ++i)
+ for (PRUint32 i = 13; i <= 15; ++i)
{
delete mProbers[i];
mProbers[i] = 0;
@@ -79,8 +82,8 @@
// disable latin2 before latin1 is available, otherwise all latin1
// will be detected as latin2 because of their similarity.
- //mProbers[10] = new nsSingleByteCharSetProber(&Latin2HungarianModel);
- //mProbers[11] = new nsSingleByteCharSetProber(&Win1250HungarianModel);
+ //mProbers[13] = new nsSingleByteCharSetProber(&Latin2HungarianModel);
+ //mProbers[14] = new nsSingleByteCharSetProber(&Win1250HungarianModel);
Reset();
}
--- nsSBCSGroupProber.h 2009/02/21 20:08:02 1.1.1.1
+++ nsSBCSGroupProber.h 2009/02/23 13:47:40 1.2
@@ -40,7 +40,7 @@
#define nsSBCSGroupProber_h__
-#define NUM_OF_SBCS_PROBERS 13
+#define NUM_OF_SBCS_PROBERS 16
class nsCharSetProber;
class nsSBCSGroupProber: public nsCharSetProber {
--- nsSBCharSetProber.h 2009/02/21 20:08:02 1.1.1.1
+++ nsSBCharSetProber.h 2009/02/23 13:47:40 1.2
@@ -118,6 +118,7 @@
extern SequenceModel Win1251BulgarianModel;
extern SequenceModel Latin2HungarianModel;
extern SequenceModel Win1250HungarianModel;
+extern SequenceModel TIS620ThaiModel;
extern SequenceModel Win1255Model;
#endif /* nsSingleByteCharSetProber_h__ */