diff --git a/intl/lwbrk/src/jisx4501class.h b/intl/lwbrk/src/jisx4501class.h
index e3032db9b21..5f6483813c6 100644
--- a/intl/lwbrk/src/jisx4501class.h
+++ b/intl/lwbrk/src/jisx4501class.h
@@ -46,7 +46,7 @@ static const PRUint32 gLBClass00[32] = {
0x7AABAAA5, // U+0020 - U+0027
0x7A7AAAA9, // U+0028 - U+002F
0x66666666, // U+0030 - U+0037
-0x1AA9AA66, // U+0038 - U+003F
+0xAAA9AA66, // U+0038 - U+003F
0x77777777, // U+0040 - U+0047
0x77777777, // U+0048 - U+004F
0x77777777, // U+0050 - U+0057
@@ -76,9 +76,9 @@ static const PRUint32 gLBClass00[32] = {
static const PRUint32 gLBClass20[32] = {
0xB5555555, // U+2000 - U+2007
0x77775555, // U+2008 - U+200F
-0x777211B1, // U+2010 - U+2017
+0x777277B7, // U+2010 - U+2017
0x77777777, // U+2018 - U+201F
-0xA2227777, // U+2020 - U+2027
+0xAAAA7777, // U+2020 - U+2027
0xB7777777, // U+2028 - U+202F
0x77744444, // U+2030 - U+2037
0x7A115107, // U+2038 - U+203F
diff --git a/intl/lwbrk/src/nsJISx4501LineBreaker.cpp b/intl/lwbrk/src/nsJISx4501LineBreaker.cpp
index b1fd3ea6fba..1e279da35ca 100644
--- a/intl/lwbrk/src/nsJISx4501LineBreaker.cpp
+++ b/intl/lwbrk/src/nsJISx4501LineBreaker.cpp
@@ -382,12 +382,14 @@ static const PRUint16 gPairConservative[MAX_CLASSES] = {
#define U_EQUAL PRUnichar('=')
#define U_PERCENT PRUnichar('%')
#define U_AMPERSAND PRUnichar('&')
+#define U_SEMICOLON PRUnichar(';')
#define U_BACKSLASH PRUnichar('\\')
#define NEED_CONTEXTUAL_ANALYSIS(c) (IS_HYPHEN(c) || \
(c) == U_SLASH || \
(c) == U_PERCENT || \
(c) == U_AMPERSAND || \
+ (c) == U_SEMICOLON || \
(c) == U_BACKSLASH)
#define IS_ASCII_DIGIT(u) (0x0030 <= (u) && (u) <= 0x0039)
@@ -431,7 +433,8 @@ IS_HYPHEN(PRUnichar u)
return (u == U_HYPHEN ||
u == 0x058A || // ARMENIAN HYPHEN
u == 0x2010 || // HYPHEN
- u == 0x2012); // FIGURE DASH
+ u == 0x2012 || // FIGURE DASH
+ u == 0x2013); // EN DASH
}
static PRInt8
@@ -608,11 +611,9 @@ public:
// is near it.
// Note that index is always larger than CONSERVATIVE_BREAK_RANGE here.
- for (PRUint32 i = index - 1; index - CONSERVATIVE_BREAK_RANGE < i; --i) {
- if (IS_NONBREAKABLE_SPACE(GetCharAt(i)))
+ for (PRUint32 i = index; index - CONSERVATIVE_BREAK_RANGE < i; --i) {
+ if (IS_NONBREAKABLE_SPACE(GetCharAt(i - 1)))
return PR_TRUE;
- if (i == 0)
- break;
}
// Note that index is always less than mLength - CONSERVATIVE_BREAK_RANGE.
for (PRUint32 i = index + 1; i < index + CONSERVATIVE_BREAK_RANGE; ++i) {
@@ -624,13 +625,9 @@ public:
PRBool HasCharacterAlready(PRUnichar aCh) {
// Be careful for the index being unsigned.
- if (mIndex == 0)
- return PR_FALSE;
- for (PRUint32 i = mIndex - 1; 0 < i; --i) {
- if (GetCharAt(i) == aCh)
+ for (PRUint32 i = mIndex; i > 0; --i) {
+ if (GetCharAt(i - 1) == aCh)
return PR_TRUE;
- if (i == 0)
- break;
}
return PR_FALSE;
}
@@ -639,14 +636,10 @@ public:
NS_ASSERTION(IS_HYPHEN(GetCharAt(mIndex)),
"current character isn't hyphen");
// Be careful for the index being unsigned.
- if (mIndex == 0)
- return PR_FALSE;
- for (PRUint32 i = mIndex - 1; 0 < i; --i) {
- PRUnichar ch = GetCharAt(i);
+ for (PRUint32 i = mIndex; i > 0; --i) {
+ PRUnichar ch = GetCharAt(i - 1);
if (!IS_HYPHEN(ch))
return ch;
- if (i == 0)
- break;
}
return U_NULL;
}
@@ -725,7 +718,7 @@ ContextualAnalysis(PRUnichar prev, PRUnichar cur, PRUnichar next,
aState.GetCharAt(aState.Index() + 3) == U_PERCENT)
return CLASS_OPEN;
}
- } else if (cur == U_AMPERSAND) {
+ } else if (cur == U_AMPERSAND || cur == U_SEMICOLON) {
// If this may be a separator of params of URL, we should break after.
if (!aState.UseConservativeBreaking(1) &&
aState.HasCharacterAlready(U_EQUAL))
diff --git a/intl/lwbrk/tools/anzx4501.html b/intl/lwbrk/tools/anzx4501.html
index 44da8ee3917..9e5f900cf43 100644
--- a/intl/lwbrk/tools/anzx4501.html
+++ b/intl/lwbrk/tools/anzx4501.html
@@ -92,10 +92,10 @@ Analysis of JIS X 4051 to Unicode General Category Mapping
| 31 |
2 |
|
-28 |
+24 |
3 |
|
-64 |
+60 |
|
|
|
@@ -112,11 +112,11 @@ Analysis of JIS X 4051 to Unicode General Category Mapping
|
|
|
-4 |
+1 |
12 |
1 |
|
-11 |
+10 |
|
|
2 |
@@ -131,10 +131,10 @@ Analysis of JIS X 4051 to Unicode General Category Mapping
|
|
|
-4 |
+1 |
|
|
-4 |
+1 |
|
|
|
@@ -155,7 +155,7 @@ Analysis of JIS X 4051 to Unicode General Category Mapping
|
|
|
-3 |
+ |
|
|
|
@@ -326,10 +326,10 @@ Analysis of JIS X 4051 to Unicode General Category Mapping
157 |
|
33 |
-54 |
+57 |
125 |
3 |
-391 |
+394 |
|
19 |
|
@@ -346,7 +346,7 @@ Analysis of JIS X 4051 to Unicode General Category Mapping
3 |
30 |
4 |
-2 |
+5 |
2 |
2 |
4 |
@@ -443,10 +443,10 @@ Analysis of JIS X 4051 to Unicode General Category Mapping
2 |
|
6 |
-21 |
+25 |
16 |
|
-46 |
+50 |
|
1 |
|
@@ -467,7 +467,7 @@ Analysis of JIS X 4051 to Unicode General Category Mapping
3 |
1 |
1 |
-16 |
+20 |
|
2 |
3 |
@@ -574,7 +574,7 @@ Analysis of JIS X 4051 to Unicode General Category Mapping
| 00 |
| |
-1 |
+ |
|
|
|
@@ -583,7 +583,7 @@ Analysis of JIS X 4051 to Unicode General Category Mapping
126 |
|
7 |
-44 |
+45 |
2 |
|
|---|
@@ -604,16 +604,16 @@ Analysis of JIS X 4051 to Unicode General Category Mapping
| 20 |
| 2 |
-11 |
-4 |
+8 |
+1 |
|
5 |
12 |
|
-101 |
+104 |
|
|
-2 |
+5 |
3 |
|
|---|
diff --git a/intl/lwbrk/tools/jisx4501class.txt b/intl/lwbrk/tools/jisx4501class.txt
index 81e1c29706b..ac1e19bee01 100644
--- a/intl/lwbrk/tools/jisx4501class.txt
+++ b/intl/lwbrk/tools/jisx4501class.txt
@@ -8,7 +8,6 @@
0021;002F;23
0030;0039;15
003C;;22
-003F;;4
003A;003F;23
0040;;18
0041;005A;18
@@ -44,15 +43,15 @@
2007;;24
2000;200B;17
200C;200F;18
-2010;;2
+2010;;18
2011;;24
-2012;2013;2
+2012;2013;18
2014;;7
2015;;18
2016;2017;18
2018;201F;18
2020;2023;18
-2024;2026;7
+2024;2026;23
2027;;23
2028;202E;18
202F;;24