ICU-9954 rbbi forward iteration performance improvement. Ensure rules always make...

X-SVN-Rev: 40461

diff --git a/icu4c/source/data/brkitr/rules/char.txt b/icu4c/source/data/brkitr/rules/char.txt
index ad070638aa381e5b59177041c60374bd59719fb7..1bfdef637a959dea8649e86be146bc514e383fd2 100644 (file)
--- a/icu4c/source/data/brkitr/rules/char.txt
+++ b/icu4c/source/data/brkitr/rules/char.txt
@@ -78,6 +78,9 @@ $Prepend [^$Control $CR $LF];
 ^$Prepend* $Regional_Indicator $Regional_Indicator / $Regional_Indicator;
 ^$Prepend* $Regional_Indicator $Regional_Indicator;
 
+# GB 999 Match a single code point if no other rule applies.
+.;
+
 ## -------------------------------------------------
 
 !!safe_reverse;

diff --git a/icu4c/source/data/brkitr/rules/line.txt b/icu4c/source/data/brkitr/rules/line.txt
index 902ca8bfea65453772c17c585a373f576fbdc953..756b19d74f69e6f85554c56363683b55ce06fb08 100644 (file)
--- a/icu4c/source/data/brkitr/rules/line.txt
+++ b/icu4c/source/data/brkitr/rules/line.txt
@@ -335,6 +335,9 @@ $RI $CM* $RI $CM* $ZWJ ($ID | $Extended_Pict | $EmojiNRK);
 # LB 30b Do not break between an Emoji Base and an Emoji Modifier
 $EB $CM* $EM;
 
+# LB 31 Break everywhere else.
+#       Match a single code point if no other rule applies.
+.;
 
 ## -------------------------------------------------
 

diff --git a/icu4c/source/data/brkitr/rules/line_fi.txt b/icu4c/source/data/brkitr/rules/line_fi.txt
index 73cc9edbd018f758196d2778d9081699e4bc0cff..5d0f666b830305918273720d38332e8edb5a6bf9 100644 (file)
--- a/icu4c/source/data/brkitr/rules/line_fi.txt
+++ b/icu4c/source/data/brkitr/rules/line_fi.txt
@@ -344,6 +344,10 @@ $RI $CM* $RI $CM* $ZWJ ($ID | $Extended_Pict | $EmojiNRK);
 # LB 30b Do not break between an Emoji Base and an Emoji Modifier
 $EB $CM* $EM;
 
+# LB 31 Break everywhere else.
+#       Match a single code point if no other rule applies.
+.;
+
 ## -------------------------------------------------
 
 !!safe_reverse;

diff --git a/icu4c/source/data/brkitr/rules/line_loose.txt b/icu4c/source/data/brkitr/rules/line_loose.txt
index 0e39e9cc3ff0e0aa6a76032036d4451a1fb7364c..ba18f1cd2c48a577b3b656c3a5054a8cb9319df3 100644 (file)
--- a/icu4c/source/data/brkitr/rules/line_loose.txt
+++ b/icu4c/source/data/brkitr/rules/line_loose.txt
@@ -347,6 +347,9 @@ $RI $CM* $RI $CM* $ZWJ ($ID | $Extended_Pict | $EmojiNRK);
 # LB 30b Do not break between an Emoji Base and an Emoji Modifier
 $EB $CM* $EM;
 
+# LB 31 Break everywhere else.
+#       Match a single code point if no other rule applies.
+.;
 
 ## -------------------------------------------------
 

diff --git a/icu4c/source/data/brkitr/rules/line_loose_cj.txt b/icu4c/source/data/brkitr/rules/line_loose_cj.txt
index 4b452de1f5bda1c035244c3c435cd2010c095e7c..48791c4d5eb0f4e375e09ad1d9b83da195a9d75d 100644 (file)
--- a/icu4c/source/data/brkitr/rules/line_loose_cj.txt
+++ b/icu4c/source/data/brkitr/rules/line_loose_cj.txt
@@ -361,6 +361,10 @@ $RI $CM* $RI $CM* $ZWJ ($ID | $Extended_Pict | $EmojiNRK);
 # LB 30b Do not break between an Emoji Base and an Emoji Modifier
 $EB $CM* $EM;
 
+# LB 31 Break everywhere else.
+#       Match a single code point if no other rule applies.
+.;
+
 ## -------------------------------------------------
 
 !!safe_reverse;

diff --git a/icu4c/source/data/brkitr/rules/line_loose_fi.txt b/icu4c/source/data/brkitr/rules/line_loose_fi.txt
index f68057bfebfbfe46efe20ca6bf84378c4d52678f..cc09db7969ddb902c5ec9403cd3da8540c6a6ef2 100644 (file)
--- a/icu4c/source/data/brkitr/rules/line_loose_fi.txt
+++ b/icu4c/source/data/brkitr/rules/line_loose_fi.txt
@@ -346,6 +346,9 @@ $RI $CM* $RI $CM* $ZWJ ($ID | $Extended_Pict | $EmojiNRK);
 # LB 30b Do not break between an Emoji Base and an Emoji Modifier
 $EB $CM* $EM;
 
+# LB 31 Break everywhere else.
+#       Match a single code point if no other rule applies.
+.;
 
 ## -------------------------------------------------
 

diff --git a/icu4c/source/data/brkitr/rules/line_normal.txt b/icu4c/source/data/brkitr/rules/line_normal.txt
index 594cb165475ed3fa30aad5f8b9c4ab5ddf7c4226..bc417cf567fd66fcdb68cb851ebbbce0778bb935 100644 (file)
--- a/icu4c/source/data/brkitr/rules/line_normal.txt
+++ b/icu4c/source/data/brkitr/rules/line_normal.txt
@@ -339,6 +339,10 @@ $RI $CM* $RI $CM* $ZWJ ($ID | $Extended_Pict | $EmojiNRK);
 # LB 30b Do not break between an Emoji Base and an Emoji Modifier
 $EB $CM* $EM;
 
+# LB 31 Break everywhere else.
+#       Match a single code point if no other rule applies.
+.;
+
 ## -------------------------------------------------
 
 !!safe_reverse;

diff --git a/icu4c/source/data/brkitr/rules/line_normal_cj.txt b/icu4c/source/data/brkitr/rules/line_normal_cj.txt
index 59c706d25624483168bf34ca5dd0b96de0c3fd59..5dbcd85ad40ac8e033e892f06273910e9a78a9f6 100644 (file)
--- a/icu4c/source/data/brkitr/rules/line_normal_cj.txt
+++ b/icu4c/source/data/brkitr/rules/line_normal_cj.txt
@@ -345,6 +345,10 @@ $RI $CM* $RI $CM* $ZWJ ($ID | $Extended_Pict | $EmojiNRK);
 # LB 30b Do not break between an Emoji Base and an Emoji Modifier
 $EB $CM* $EM;
 
+# LB 31 Break everywhere else.
+#       Match a single code point if no other rule applies.
+.;
+
 ## -------------------------------------------------
 
 !!safe_reverse;

diff --git a/icu4c/source/data/brkitr/rules/line_normal_fi.txt b/icu4c/source/data/brkitr/rules/line_normal_fi.txt
index b6e10b90be48f7159b1ed893cdd3ba088592273e..5cd5553605f6485697104ddc5c5f2dcd1dbf9b1f 100644 (file)
--- a/icu4c/source/data/brkitr/rules/line_normal_fi.txt
+++ b/icu4c/source/data/brkitr/rules/line_normal_fi.txt
@@ -342,6 +342,10 @@ $RI $CM* $RI $CM* $ZWJ ($ID | $Extended_Pict | $EmojiNRK);
 # LB 30b Do not break between an Emoji Base and an Emoji Modifier
 $EB $CM* $EM;
 
+# LB 31 Break everywhere else.
+#       Match a single code point if no other rule applies.
+.;
+
 ## -------------------------------------------------
 
 !!safe_reverse;

diff --git a/icu4c/source/data/brkitr/rules/word.txt b/icu4c/source/data/brkitr/rules/word.txt
index 7c4e9a39a64f140141951db9bcef040e39d40889..617205debc26ef868974112cbaa5c027f3e9773a 100644 (file)
--- a/icu4c/source/data/brkitr/rules/word.txt
+++ b/icu4c/source/data/brkitr/rules/word.txt
@@ -194,6 +194,10 @@ $ExtendNumLetEx  $KatakanaEx     {400};    #  (13b)
 $HangulSyllable $HangulSyllable {200};
 $KanaKanji $KanaKanji {400}; # different rule status if both kana and kanji found 
 
+# Rule 999
+#     Match a single code point if no other rule applies.
+.;
+
 
 ## -------------------------------------------------
 

diff --git a/icu4c/source/data/brkitr/rules/word_POSIX.txt b/icu4c/source/data/brkitr/rules/word_POSIX.txt
index ec46da6291540c139d0a726cab6283e6da2f6df7..5ea6a05ce4b969a7df581a099c2d412291eee9d5 100644 (file)
--- a/icu4c/source/data/brkitr/rules/word_POSIX.txt
+++ b/icu4c/source/data/brkitr/rules/word_POSIX.txt
@@ -194,6 +194,10 @@ $ExtendNumLetEx  $KatakanaEx     {400};    #  (13b)
 $HangulSyllable $HangulSyllable {200};
 $KanaKanji $KanaKanji {400}; # different rule status if both kana and kanji found
 
+# Rule 999
+#     Match a single code point if no other rule applies.
+.;
+
 
 ## -------------------------------------------------