From e220fe9dd65ca58ae2a3bbb38ded46fba01d38fb Mon Sep 17 00:00:00 2001
From: Andy Heninger <andy.heninger@gmail.com>
Date: Tue, 26 Sep 2017 20:16:03 +0000
Subject: [PATCH] ICU-9954 rbbi forward iteration performance improvement.
 Ensure rules always make some progress; keep engine out of the fail-safe fall
 back path for rules that do not advance.

X-SVN-Rev: 40461
---
 icu4c/source/data/brkitr/rules/char.txt           | 3 +++
 icu4c/source/data/brkitr/rules/line.txt           | 3 +++
 icu4c/source/data/brkitr/rules/line_fi.txt        | 4 ++++
 icu4c/source/data/brkitr/rules/line_loose.txt     | 3 +++
 icu4c/source/data/brkitr/rules/line_loose_cj.txt  | 4 ++++
 icu4c/source/data/brkitr/rules/line_loose_fi.txt  | 3 +++
 icu4c/source/data/brkitr/rules/line_normal.txt    | 4 ++++
 icu4c/source/data/brkitr/rules/line_normal_cj.txt | 4 ++++
 icu4c/source/data/brkitr/rules/line_normal_fi.txt | 4 ++++
 icu4c/source/data/brkitr/rules/word.txt           | 4 ++++
 icu4c/source/data/brkitr/rules/word_POSIX.txt     | 4 ++++
 11 files changed, 40 insertions(+)

diff --git a/icu4c/source/data/brkitr/rules/char.txt b/icu4c/source/data/brkitr/rules/char.txt
index ad070638aa3..1bfdef637a9 100644
--- 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 902ca8bfea6..756b19d74f6 100644
--- 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 73cc9edbd01..5d0f666b830 100644
--- 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 0e39e9cc3ff..ba18f1cd2c4 100644
--- 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 4b452de1f5b..48791c4d5eb 100644
--- 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 f68057bfebf..cc09db7969d 100644
--- 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 594cb165475..bc417cf567f 100644
--- 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 59c706d2562..5dbcd85ad40 100644
--- 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 b6e10b90be4..5cd5553605f 100644
--- 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 7c4e9a39a64..617205debc2 100644
--- 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 ec46da62915..5ea6a05ce4b 100644
--- 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.
+.;
+
 
 ## -------------------------------------------------
 
-- 
2.40.0