From 6c1e41e0f20efe32cf19f4dba0d93c610627db9d Mon Sep 17 00:00:00 2001
From: Markus Scherer <markus.icu@gmail.com>
Date: Wed, 7 Jun 2017 07:13:34 +0000
Subject: [PATCH] ICU-13197 catch one more edge case where a mapping has no
 compose boundary after it

X-SVN-Rev: 40156
---
 icu4c/source/data/in/nfkc.nrm             | Bin 54160 -> 54160 bytes
 icu4c/source/data/in/nfkc_cf.nrm          | Bin 51864 -> 51864 bytes
 icu4c/source/data/in/uts46.nrm            | Bin 61060 -> 61060 bytes
 icu4c/source/tools/gennorm2/n2builder.cpp |  28 +++++++++++++---------
 icu4c/source/tools/gennorm2/n2builder.h   |   2 --
 icu4c/source/tools/gennorm2/norms.cpp     |   2 +-
 icu4c/source/tools/gennorm2/norms.h       |   4 +++-
 7 files changed, 21 insertions(+), 15 deletions(-)

diff --git a/icu4c/source/data/in/nfkc.nrm b/icu4c/source/data/in/nfkc.nrm
index 8f6802b9c81066a91d1f72f8e342b4eba2f7dfbd..c4b7e7228448588f615cff3e7f19b6f7e973b575 100644
GIT binary patch
delta 16
YcmbQRoO!}><_#D2F)D4oxNpA*071$J>i_@%

delta 16
YcmbQRoO!}><_#D2F*0qwxNpA*06_`|%K!iX

diff --git a/icu4c/source/data/in/nfkc_cf.nrm b/icu4c/source/data/in/nfkc_cf.nrm
index 29abf1330f44d7c266b8b82996e75e1b5a061d88..9d91ad32ab40ca5b42b683da21a458e8064b6eb2 100644
GIT binary patch
delta 16
YcmbO+m3hWg<_&o}8I?BY@6;3o06J_2Z~y=R

delta 16
YcmbO+m3hWg<_&o}8JRZc@6;3o06DA%Pyhe`

diff --git a/icu4c/source/data/in/uts46.nrm b/icu4c/source/data/in/uts46.nrm
index e3f0dca90c93d45e8a78b84ceccf0858a6b3f7aa..7270a67b41d09430c64d307e1a10207eb484a4f8 100644
GIT binary patch
delta 16
XcmZp<%iMC8c|+A*My1WwcR2+CL8u29

delta 16
XcmZp<%iMC8c|+A*MyAcxcR2+CK<Ebe

diff --git a/icu4c/source/tools/gennorm2/n2builder.cpp b/icu4c/source/tools/gennorm2/n2builder.cpp
index 587f0dc9f46..374e91c4884 100644
--- a/icu4c/source/tools/gennorm2/n2builder.cpp
+++ b/icu4c/source/tools/gennorm2/n2builder.cpp
@@ -208,7 +208,7 @@ void Normalizer2DataBuilder::removeMapping(UChar32 c) {
 
 UBool Normalizer2DataBuilder::hasNoCompBoundaryAfter(const BuilderReorderingBuffer &buffer) {
     if(buffer.isEmpty()) {
-        return TRUE;  // maps-to-empty-string is no boundary of any kind
+        return TRUE;  // Maps-to-empty-string is no boundary of any kind.
     }
     int32_t lastStarterIndex=buffer.lastStarterIndex();
     if(lastStarterIndex<0) {
@@ -223,31 +223,37 @@ UBool Normalizer2DataBuilder::hasNoCompBoundaryAfter(const BuilderReorderingBuff
         return lastStarterIndex==buffer.length()-1;
     }
     // Note: There can be no Hangul syllable in the fully decomposed mapping.
-    const Norm *starterNorm=&norms.getNormRef(starter);
-    if(starterNorm->compositions==NULL) {
-        return FALSE;  // the last starter does not combine forward
+    const Norm *starterNorm=norms.getNorm(starter);
+    if(starterNorm==nullptr || starterNorm->compositions==nullptr) {
+        return FALSE;  // The last starter does not combine forward.
     }
     // Compose as far as possible, and see if further compositions are possible.
     uint8_t prevCC=0;
     for(int32_t combMarkIndex=lastStarterIndex+1; combMarkIndex<buffer.length(); ++combMarkIndex) {
         uint8_t cc=buffer.ccAt(combMarkIndex);  // !=0 because after last starter
         if(norms.combinesWithCCBetween(*starterNorm, prevCC, cc)) {
+            // The starter combines with a mark that reorders before the current one.
             return TRUE;
         }
         if(prevCC<cc && (starter=starterNorm->combine(buffer.charAt(combMarkIndex)))>=0) {
-            starterNorm=&norms.getNormRef(starter);
-            if(starterNorm->compositions==NULL) {
-                return FALSE;  // the composite does not combine further
+            // The starter combines with this mark into a composite replacement starter.
+            starterNorm=norms.getNorm(starter);
+            if(starterNorm==nullptr || starterNorm->compositions==nullptr) {
+                return FALSE;  // The composite does not combine further.
             }
             // Keep prevCC because we "removed" the combining mark.
         } else {
             prevCC=cc;
         }
     }
-    // TRUE if the final, forward-combining starter is at the end.
-    return prevCC==0;
-    // TODO?! prevCC==0 || norms.combinesWithCCBetween(*starterNorm, prevCC, int32_t! 0x100)
-    // TODO?! actually, should check if it combines with any cc not seen here
+    if(prevCC==0) {
+        return TRUE;  // forward-combining starter at the very end
+    }
+    if(norms.combinesWithCCBetween(*starterNorm, prevCC, 256)) {
+        // The starter combines with another mark.
+        return TRUE;
+    }
+    return FALSE;
 }
 
 void Normalizer2DataBuilder::postProcess(Norm &norm) {
diff --git a/icu4c/source/tools/gennorm2/n2builder.h b/icu4c/source/tools/gennorm2/n2builder.h
index 130a0bd6e6e..1fca0b70571 100644
--- a/icu4c/source/tools/gennorm2/n2builder.h
+++ b/icu4c/source/tools/gennorm2/n2builder.h
@@ -77,8 +77,6 @@ private:
      * (especially for a "YesNo" which has a round-trip mapping).
      * This flag is used in Normalizer2Impl::hasCompBoundaryAfter().
      *
-     * Modifies the buffer (partially composes it).
-     *
      * A starter character with a mapping does not have a composition boundary after it
      * if the character itself combines-forward (which is tested by the caller of this function),
      * or it is deleted (mapped to the empty string),
diff --git a/icu4c/source/tools/gennorm2/norms.cpp b/icu4c/source/tools/gennorm2/norms.cpp
index 7ea93778d89..8dba9c81558 100644
--- a/icu4c/source/tools/gennorm2/norms.cpp
+++ b/icu4c/source/tools/gennorm2/norms.cpp
@@ -131,7 +131,7 @@ void Norms::reorder(UnicodeString &mapping, BuilderReorderingBuffer &buffer) con
     }
 }
 
-UBool Norms::combinesWithCCBetween(const Norm &norm, uint8_t lowCC, uint8_t highCC) const {
+UBool Norms::combinesWithCCBetween(const Norm &norm, uint8_t lowCC, int32_t highCC) const {
     if((highCC-lowCC)>=2) {
         int32_t length;
         const CompositionPair *pairs=norm.getCompositionPairs(length);
diff --git a/icu4c/source/tools/gennorm2/norms.h b/icu4c/source/tools/gennorm2/norms.h
index a7ee4974b0d..052298ed7ed 100644
--- a/icu4c/source/tools/gennorm2/norms.h
+++ b/icu4c/source/tools/gennorm2/norms.h
@@ -156,7 +156,9 @@ public:
     uint8_t getCC(UChar32 c) const { return getNormRef(c).cc; }
 
     void reorder(UnicodeString &mapping, BuilderReorderingBuffer &buffer) const;
-    UBool combinesWithCCBetween(const Norm &norm, uint8_t lowCC, uint8_t highCC) const;
+
+    // int32_t highCC not uint8_t so that we can pass in 256 as the upper limit.
+    UBool combinesWithCCBetween(const Norm &norm, uint8_t lowCC, int32_t highCC) const;
 
     class Enumerator {
     public: