From f247f08e370626bbb427acd8f4a400fd875350a3 Mon Sep 17 00:00:00 2001

Message-Id: <f247f08e370626bbb427acd8f4a400fd875350a3.1531842113.git.erack@redhat.com>

From: Eike Rathke <erack@redhat.com>

Date: Tue, 17 Apr 2018 20:13:52 +0200

Subject: [PATCH] Upgrade to ICU 61.1

MIME-Version: 1.0

Content-Type: multipart/mixed; boundary="------------erAck-patch-parts"

This is a multi-part message in MIME format.

--------------erAck-patch-parts

Content-Type: text/plain; charset=UTF-8; format=fixed

Content-Transfer-Encoding: 8bit

Change-Id: I89c1c3d13d85decc72576744de2a16d20471d29d

Reviewed-on: https://gerrit.libreoffice.org/53064

Tested-by: Jenkins <ci@libreoffice.org>

Reviewed-by: Eike Rathke <erack@redhat.com>

---

 configure.ac                                  |   4 +-

 download.lst                                  |   8 +-

 external/icu/UnpackedTarball_icu.mk           |   3 +-

 ...patch.1 => icu4c-61-werror-shadow.patch.1} |  30 ++-

 external/icu/icu4c-khmerbreakengine.patch.1   | 246 +++++++++---------

 external/icu/icu4c-ubsan.patch.1              |  91 -------

 6 files changed, 147 insertions(+), 235 deletions(-)

 rename external/icu/{icu4c-59-werror-shadow.patch.1 => icu4c-61-werror-shadow.patch.1} (50%)

 delete mode 100644 external/icu/icu4c-ubsan.patch.1

--------------erAck-patch-parts

Content-Type: text/x-patch; name="0001-Upgrade-to-ICU-61.1.patch"

Content-Transfer-Encoding: 8bit

Content-Disposition: attachment; filename="0001-Upgrade-to-ICU-61.1.patch"

diff --git a/configure.ac b/configure.ac

index bdc374699cb9..458f9f3c2735 100644

--- a/configure.ac

+++ b/configure.ac

@@ -9001,8 +9001,8 @@ SYSTEM_GENBRK=

 SYSTEM_GENCCODE=

 SYSTEM_GENCMN=

-ICU_MAJOR=60

-ICU_MINOR=2

+ICU_MAJOR=61

+ICU_MINOR=1

 ICU_RECLASSIFIED_PREPEND_SET_EMPTY="TRUE"

 ICU_RECLASSIFIED_CONDITIONAL_JAPANESE_STARTER="TRUE"

 ICU_RECLASSIFIED_HEBREW_LETTER="TRUE"

diff --git a/download.lst b/download.lst

index 0a1fa65396d1..76f067bff4d7 100644

--- a/download.lst

+++ b/download.lst

@@ -102,10 +102,10 @@ export HUNSPELL_SHA256SUM := 3cd9ceb062fe5814f668e4f22b2fa6e3ba0b339b921739541ce

 export HUNSPELL_TARBALL := hunspell-1.6.2.tar.gz

 export HYPHEN_SHA256SUM := 304636d4eccd81a14b6914d07b84c79ebb815288c76fe027b9ebff6ff24d5705

 export HYPHEN_TARBALL := 5ade6ae2a99bc1e9e57031ca88d36dad-hyphen-2.8.8.tar.gz

-export ICU_SHA256SUM := f073ea8f35b926d70bb33e6577508aa642a8b316a803f11be20af384811db418

-export ICU_TARBALL := icu4c-60_2-src.tgz

-export ICU_DATA_SHA256SUM := 68f42ad0c9e0a5a5af8eba0577ba100833912288bad6e4d1f42ff480bbcfd4a9

-export ICU_DATA_TARBALL := icu4c-60_2-data.zip

+export ICU_SHA256SUM := d007f89ae8a2543a53525c74359b65b36412fa84b3349f1400be6dcf409fafef

+export ICU_TARBALL := icu4c-61_1-src.tgz

+export ICU_DATA_SHA256SUM := d149ed0985b5a6e16a9d8ed66f105dd58fd334c276779f74241cfa656ed2830a

+export ICU_DATA_TARBALL := icu4c-61_1-data.zip

 export JFREEREPORT_FLOW_ENGINE_SHA256SUM := 233f66e8d25c5dd971716d4200203a612a407649686ef3b52075d04b4c9df0dd

 export JFREEREPORT_FLOW_ENGINE_TARBALL := ba2930200c9f019c2d93a8c88c651a0f-flow-engine-0.9.4.zip

 export JFREEREPORT_FLUTE_SHA256SUM := 1b5b24f7bc543c0362b667692f78db8bab4ed6dafc6172f104d0bd3757d8a133

diff --git a/external/icu/UnpackedTarball_icu.mk b/external/icu/UnpackedTarball_icu.mk

index a4d0b16ecb36..b81cdaab6242 100644

--- a/external/icu/UnpackedTarball_icu.mk

+++ b/external/icu/UnpackedTarball_icu.mk

@@ -27,15 +27,14 @@ $(eval $(call gb_UnpackedTarball_add_patches,icu,\

 	external/icu/icu4c-solarisgcc.patch.1 \

 	external/icu/icu4c-mkdir.patch.1 \

 	external/icu/icu4c-$(if $(filter ANDROID,$(OS)),android,rpath).patch.1 \

-	external/icu/icu4c-ubsan.patch.1 \

 	external/icu/icu4c-icu11100.patch.1 \

 	external/icu/icu4c-scriptrun.patch.1 \

 	external/icu/icu4c-rtti.patch.1 \

 	external/icu/icu4c-clang-cl.patch.1 \

 	$(if $(filter-out ANDROID,$(OS)),external/icu/icu4c-icudata-stdlibs.patch.1) \

 	external/icu/icu4c-khmerbreakengine.patch.1 \

-	external/icu/icu4c-59-werror-shadow.patch.1 \

 	external/icu/ofz4860.patch.2 \

+	external/icu/icu4c-61-werror-shadow.patch.1 \

 ))

 $(eval $(call gb_UnpackedTarball_add_file,icu,source/data/brkitr/khmerdict.dict,external/icu/khmerdict.dict))

diff --git a/external/icu/icu4c-59-werror-shadow.patch.1 b/external/icu/icu4c-61-werror-shadow.patch.1

similarity index 50%

rename from external/icu/icu4c-59-werror-shadow.patch.1

rename to external/icu/icu4c-61-werror-shadow.patch.1

index fb88244aff13..b00f76317eff 100644

--- a/external/icu/icu4c-59-werror-shadow.patch.1

+++ b/external/icu/icu4c-61-werror-shadow.patch.1

@@ -1,33 +1,35 @@

+# https://ssl.icu-project.org/trac/ticket/13709

+# Werror=shadow fails for unistr.h

 diff -ur icu.org/source/common/unicode/unistr.h icu/source/common/unicode/unistr.h

---- icu.org/source/common/unicode/unistr.h	2017-03-29 06:44:37.000000000 +0200

-+++ icu/source/common/unicode/unistr.h	2017-04-24 11:59:51.782076511 +0200

-@@ -3080,11 +3080,11 @@

+--- icu.org/source/common/unicode/unistr.h	2018-03-26 15:38:29.000000000 +0200

++++ icu/source/common/unicode/unistr.h	2018-04-18 10:44:16.321188314 +0200

+@@ -3053,11 +3053,11 @@

     * uint16_t * constructor.

     * Delegates to UnicodeString(const char16_t *, int32_t).

     * @param text UTF-16 string

 -   * @param length string length

-+   * @param length_ string length

-    * @draft ICU 59

++   * @param textLength string length

+    * @stable ICU 59

     */

 -  UnicodeString(const uint16_t *text, int32_t length) :

 -      UnicodeString(ConstChar16Ptr(text), length) {}

-+  UnicodeString(const uint16_t *text, int32_t length_) :

-+      UnicodeString(ConstChar16Ptr(text), length_) {}

++  UnicodeString(const uint16_t *text, int32_t textLength) :

++      UnicodeString(ConstChar16Ptr(text), textLength) {}

  #endif

-   /*

-@@ -3097,11 +3097,11 @@

+ #if U_SIZEOF_WCHAR_T==2 || defined(U_IN_DOXYGEN)

+@@ -3066,11 +3066,11 @@

     * (Only defined if U_SIZEOF_WCHAR_T==2.)

     * Delegates to UnicodeString(const char16_t *, int32_t).

     * @param text NUL-terminated UTF-16 string

 -   * @param length string length

-+   * @param length_ string length

-    * @draft ICU 59

++   * @param textLength string length

+    * @stable ICU 59

     */

 -  UnicodeString(const wchar_t *text, int32_t length) :

 -      UnicodeString(ConstChar16Ptr(text), length) {}

-+  UnicodeString(const wchar_t *text, int32_t length_) :

-+      UnicodeString(ConstChar16Ptr(text), length_) {}

++  UnicodeString(const wchar_t *text, int32_t textLength) :

++      UnicodeString(ConstChar16Ptr(text), textLength) {}

  #endif

-   /*

+   /**

diff --git a/external/icu/icu4c-khmerbreakengine.patch.1 b/external/icu/icu4c-khmerbreakengine.patch.1

index 6b45b3743611..9f134dd961b1 100644

--- a/external/icu/icu4c-khmerbreakengine.patch.1

+++ b/external/icu/icu4c-khmerbreakengine.patch.1

@@ -1,16 +1,18 @@

 diff -ur icu.org/source/common/dictbe.cpp icu/source/common/dictbe.cpp

---- icu.org/source/common/dictbe.cpp	2017-01-20 01:20:31.000000000 +0100

-+++ icu/source/common/dictbe.cpp	2017-04-21 23:14:23.845894374 +0200

-@@ -29,8 +29,17 @@

+--- icu.org/source/common/dictbe.cpp	2018-03-26 15:38:30.000000000 +0200

++++ icu/source/common/dictbe.cpp	2018-04-17 17:55:38.620944919 +0200

+@@ -29,7 +29,19 @@

   ******************************************************************

   */

--DictionaryBreakEngine::DictionaryBreakEngine(uint32_t breakTypes) {

-+DictionaryBreakEngine::DictionaryBreakEngine(uint32_t breakTypes) :

-+    clusterLimit(3)

-+{

+-DictionaryBreakEngine::DictionaryBreakEngine() {

++DictionaryBreakEngine::DictionaryBreakEngine()

++    : fTypes(0), clusterLimit(0) {

++}

++

++DictionaryBreakEngine::DictionaryBreakEngine(uint32_t breakTypes)

++    : fTypes(breakTypes), clusterLimit(3) {

 +    UErrorCode status = U_ZERO_ERROR;

-     fTypes = breakTypes;

 +    fViramaSet.applyPattern(UNICODE_STRING_SIMPLE("[[:ccc=VR:]]"), status);

 +

 +    // note Skip Sets contain fIgnoreSet characters too.

@@ -20,16 +22,7 @@ diff -ur icu.org/source/common/dictbe.cpp icu/source/common/dictbe.cpp

  }

  DictionaryBreakEngine::~DictionaryBreakEngine() {

-@@ -92,7 +101,7 @@

-         result = divideUpDictionaryRange(text, rangeStart, rangeEnd, foundBreaks);

-         utext_setNativeIndex(text, current);

-     }

--    

-+

-     return result;

- }

- 

-@@ -103,6 +112,169 @@

+@@ -76,6 +88,169 @@

      fSet.compact();

  }

@@ -199,7 +192,7 @@ diff -ur icu.org/source/common/dictbe.cpp icu/source/common/dictbe.cpp

  /*

   ******************************************************************

   * PossibleWord

-@@ -130,35 +302,35 @@

+@@ -103,35 +278,35 @@

  public:

      PossibleWord() : count(0), prefix(0), offset(-1), mark(0), current(0) {};

      ~PossibleWord() {};

@@ -244,7 +237,7 @@ diff -ur icu.org/source/common/dictbe.cpp icu/source/common/dictbe.cpp

          // Dictionary leaves text after longest prefix, not longest word. Back up.

          if (count <= 0) {

              utext_setNativeIndex(text, start);

-@@ -830,51 +1002,28 @@

+@@ -803,51 +978,28 @@

   * KhmerBreakEngine

   */

@@ -265,7 +258,8 @@ diff -ur icu.org/source/common/dictbe.cpp icu/source/common/dictbe.cpp

 -static const int32_t KHMER_MIN_WORD_SPAN = KHMER_MIN_WORD * 2;

 -

  KhmerBreakEngine::KhmerBreakEngine(DictionaryMatcher *adoptDictionary, UErrorCode &status)

-     : DictionaryBreakEngine((1 << UBRK_WORD) | (1 << UBRK_LINE)),

+-    : DictionaryBreakEngine(),

++    : DictionaryBreakEngine((1 << UBRK_WORD) | (1 << UBRK_LINE)),

        fDictionary(adoptDictionary)

  {

 -    fKhmerWordSet.applyPattern(UNICODE_STRING_SIMPLE("[[:Khmr:]&[:LineBreak=SA:]]"), status);

@@ -301,13 +295,13 @@ diff -ur icu.org/source/common/dictbe.cpp icu/source/common/dictbe.cpp

 -    fEndWordSet.compact();

 -    fBeginWordSet.compact();

 -//    fSuffixSet.compact();

-+	fIgnoreSet.compact();

-+	fBaseSet.compact();

-+	fPuncSet.compact();

++    fIgnoreSet.compact();

++    fBaseSet.compact();

++    fPuncSet.compact();

  }

  KhmerBreakEngine::~KhmerBreakEngine() {

-@@ -886,180 +1035,204 @@

+@@ -859,180 +1011,204 @@

                                                  int32_t rangeStart,

                                                  int32_t rangeEnd,

                                                  UVector32 &foundBreaks ) const {

@@ -350,17 +344,6 @@ diff -ur icu.org/source/common/dictbe.cpp icu/source/common/dictbe.cpp

 +            foundBreaks.push(rangeEnd, status);

 +        return foundBreaks.size() - wordsFound;

 +    }

-+

-+    scanStart = rangeStart;

-+    scanWJ(text, scanStart, rangeEnd, before, after);

-+    if (startZwsp || initAfter >= before) {

-+        after = initAfter;

-+        before = 0;

-+    }

-+    if (!endZwsp && after > finalBefore && after < rangeEnd)

-+        endZwsp = true;

-+    if (endZwsp && before > finalBefore)

-+        before = finalBefore;

 -    while (U_SUCCESS(status) && (current = (int32_t)utext_getNativeIndex(text)) < rangeEnd) {

 -        cuWordLength = 0;

@@ -375,7 +358,17 @@ diff -ur icu.org/source/common/dictbe.cpp icu/source/common/dictbe.cpp

 -            cpWordLength = words[wordsFound % KHMER_LOOKAHEAD].markedCPLength();

 -            wordsFound += 1;

 -        }

--

++    scanStart = rangeStart;

++    scanWJ(text, scanStart, rangeEnd, before, after);

++    if (startZwsp || initAfter >= before) {

++        after = initAfter;

++        before = 0;

++    }

++    if (!endZwsp && after > finalBefore && after < rangeEnd)

++        endZwsp = true;

++    if (endZwsp && before > finalBefore)

++        before = finalBefore;

+ 

 -        // If there was more than one, see which one can take us forward the most words

 -        else if (candidates > 1) {

 -            // If we're already at the end of the range, we're done

@@ -390,22 +383,6 @@ diff -ur icu.org/source/common/dictbe.cpp icu/source/common/dictbe.cpp

 -                        words[wordsFound % KHMER_LOOKAHEAD].markCurrent();

 -                        wordsMatched = 2;

 -                    }

--

--                    // If we're already at the end of the range, we're done

--                    if ((int32_t)utext_getNativeIndex(text) >= rangeEnd) {

--                        goto foundBest;

--                    }

--

--                    // See if any of the possible second words is followed by a third word

--                    do {

--                        // If we find a third word, stop right away

--                        if (words[(wordsFound + 2) % KHMER_LOOKAHEAD].candidates(text, fDictionary, rangeEnd)) {

--                            words[wordsFound % KHMER_LOOKAHEAD].markCurrent();

--                            goto foundBest;

--                        }

--                    }

--                    while (words[(wordsFound + 1) % KHMER_LOOKAHEAD].backUp(text));

--                }

 +    utext_setNativeIndex(text, rangeStart);

 +    int32_t numCodePts = rangeEnd - rangeStart;

 +    // bestSnlp[i] is the snlp of the best segmentation of the first i

@@ -415,7 +392,11 @@ diff -ur icu.org/source/common/dictbe.cpp icu/source/common/dictbe.cpp

 +    for(int32_t i = 1; i <= numCodePts; i++) {

 +        bestSnlp.addElement(kuint32max, status);

 +    }

-+

+ 

+-                    // If we're already at the end of the range, we're done

+-                    if ((int32_t)utext_getNativeIndex(text) >= rangeEnd) {

+-                        goto foundBest;

+-                    }

 +    // prev[i] is the index of the last code point in the previous word in

 +    // the best segmentation of the first i characters. Note negative implies

 +	// that the code point is part of an unknown word.

@@ -423,7 +404,17 @@ diff -ur icu.org/source/common/dictbe.cpp icu/source/common/dictbe.cpp

 +    for(int32_t i = 0; i <= numCodePts; i++) {

 +        prev.addElement(kuint32max, status);

 +    }

-+

+ 

+-                    // See if any of the possible second words is followed by a third word

+-                    do {

+-                        // If we find a third word, stop right away

+-                        if (words[(wordsFound + 2) % KHMER_LOOKAHEAD].candidates(text, fDictionary, rangeEnd)) {

+-                            words[wordsFound % KHMER_LOOKAHEAD].markCurrent();

+-                            goto foundBest;

+-                        }

+-                    }

+-                    while (words[(wordsFound + 1) % KHMER_LOOKAHEAD].backUp(text));

+-                }

 +    const int32_t maxWordSize = 20;

 +    UVector32 values(maxWordSize, status);

 +    values.setSize(maxWordSize);

@@ -528,27 +519,17 @@ diff -ur icu.org/source/common/dictbe.cpp icu/source/common/dictbe.cpp

 -                if (cuWordLength <= 0) {

 -                    wordsFound += 1;

 -                }

--

--                // Update the length with the passed-over characters

--                cuWordLength += chars;

--            }

--            else {

--                // Back up to where we were for next iteration

--                utext_setNativeIndex(text, current+cuWordLength);

 +                } while (fMarkSet.contains(c) || fIgnoreSet.contains(c));

 +                values.setElementAt(BADSNLP, count);

 +                lengths.setElementAt(utext_getNativeIndex(text) - currix, count++);

 +            } else {

 +                values.setElementAt(BADSNLP, count);

 +                lengths.setElementAt(1, count++);

-             }

-         }

++            }

++        }

--        // Never stop before a combining mark.

--        int32_t currPos;

--        while ((currPos = (int32_t)utext_getNativeIndex(text)) < rangeEnd && fMarkSet.contains(utext_current32(text))) {

--            utext_next32(text);

--            cuWordLength += (int32_t)utext_getNativeIndex(text) - currPos;

+-                // Update the length with the passed-over characters

+-                cuWordLength += chars;

 +        for (int32_t j = 0; j < count; j++) {

 +            uint32_t v = values.elementAti(j);

 +            int32_t newSnlp = bestSnlp.elementAti(i) + v;

@@ -559,7 +540,10 @@ diff -ur icu.org/source/common/dictbe.cpp icu/source/common/dictbe.cpp

 +                ++ln;

 +                utext_next32(text);

 +                c = utext_current32(text);

-+            }

+             }

+-            else {

+-                // Back up to where we were for next iteration

+-                utext_setNativeIndex(text, current+cuWordLength);

 +            int32_t ln_j_i = ln + i;   // yes really i!

 +            if (newSnlp < bestSnlp.elementAti(ln_j_i)) {

 +                if (v == BADSNLP) {

@@ -572,9 +556,38 @@ diff -ur icu.org/source/common/dictbe.cpp icu/source/common/dictbe.cpp

 +                else

 +                    prev.setElementAt(i, ln_j_i);

 +                bestSnlp.setElementAt(newSnlp, ln_j_i);

-+            }

+             }

          }

 -

+-        // Never stop before a combining mark.

+-        int32_t currPos;

+-        while ((currPos = (int32_t)utext_getNativeIndex(text)) < rangeEnd && fMarkSet.contains(utext_current32(text))) {

+-            utext_next32(text);

+-            cuWordLength += (int32_t)utext_getNativeIndex(text) - currPos;

++    }

++    // Start pushing the optimal offset index into t_boundary (t for tentative).

++    // prev[numCodePts] is guaranteed to be meaningful.

++    // We'll first push in the reverse order, i.e.,

++    // t_boundary[0] = numCodePts, and afterwards do a swap.

++    UVector32 t_boundary(numCodePts+1, status);

++

++    int32_t numBreaks = 0;

++    // No segmentation found, set boundary to end of range

++    while (numCodePts >= 0 && (uint32_t)bestSnlp.elementAti(numCodePts) == kuint32max) {

++        --numCodePts;

++    }

++    if (numCodePts < 0) {

++        t_boundary.addElement(numCodePts, status);

++        numBreaks++;

++    } else {

++        for (int32_t i = numCodePts; (uint32_t)i != kuint32max; i = prev.elementAti(i)) {

++            if (i < 0) i = -i;

++            t_boundary.addElement(i, status);

++            numBreaks++;

+         }

++        U_ASSERT(prev.elementAti(t_boundary.elementAti(numBreaks - 1)) == 0);

++    }

+ 

 -        // Look ahead for possible suffixes if a dictionary word does not follow.

 -        // We do this in code rather than using a rule so that the heuristic

 -        // resynch continues to function. For example, one of the suffix characters

@@ -616,30 +629,6 @@ diff -ur icu.org/source/common/dictbe.cpp icu/source/common/dictbe.cpp

 -        // Did we find a word on this iteration? If so, push it on the break stack

 -        if (cuWordLength > 0) {

 -            foundBreaks.push((current+cuWordLength), status);

-+    }

-+    // Start pushing the optimal offset index into t_boundary (t for tentative).

-+    // prev[numCodePts] is guaranteed to be meaningful.

-+    // We'll first push in the reverse order, i.e.,

-+    // t_boundary[0] = numCodePts, and afterwards do a swap.

-+    UVector32 t_boundary(numCodePts+1, status);

-+

-+    int32_t numBreaks = 0;

-+    // No segmentation found, set boundary to end of range

-+    while (numCodePts >= 0 && (uint32_t)bestSnlp.elementAti(numCodePts) == kuint32max) {

-+        --numCodePts;

-+    }

-+    if (numCodePts < 0) {

-+        t_boundary.addElement(numCodePts, status);

-+        numBreaks++;

-+    } else {

-+        for (int32_t i = numCodePts; (uint32_t)i != kuint32max; i = prev.elementAti(i)) {

-+            if (i < 0) i = -i;

-+            t_boundary.addElement(i, status);

-+            numBreaks++;

-+        }

-+        U_ASSERT(prev.elementAti(t_boundary.elementAti(numBreaks - 1)) == 0);

-+    }

-+

 +    // Now that we're done, convert positions in t_boundary[] (indices in

 +    // the normalized input string) back to indices in the original input UText

 +    // while reversing t_boundary and pushing values to foundBreaks.

@@ -669,38 +658,35 @@ diff -ur icu.org/source/common/dictbe.cpp icu/source/common/dictbe.cpp

  #if !UCONFIG_NO_NORMALIZATION

 diff -ur icu.org/source/common/dictbe.h icu/source/common/dictbe.h

---- icu.org/source/common/dictbe.h	2017-01-20 01:20:31.000000000 +0100

-+++ icu/source/common/dictbe.h	2017-04-21 23:14:23.845894374 +0200

-@@ -34,6 +34,15 @@

+--- icu.org/source/common/dictbe.h	2018-03-26 15:38:30.000000000 +0200

++++ icu/source/common/dictbe.h	2018-04-17 14:55:33.307639865 +0200

+@@ -34,7 +34,8 @@

+  * threads without synchronization.

   */

  class DictionaryBreakEngine : public LanguageBreakEngine {

-  private:

-+

-+  /**

-+   * 
Default constructor.

-+   *

-+   */

-+  DictionaryBreakEngine();

-+

+- private:

 + protected:

 +

      /**

       * The set of characters handled by this engine

       * @internal

-@@ -48,11 +57,63 @@

+@@ -42,14 +43,84 @@

-   uint32_t      fTypes;

+   UnicodeSet    fSet;

 +  const int32_t WJ   = 0x2060;

 +  const int32_t ZWSP = 0x200B;

 +

-   /**

--   * 
Default constructor.

--   *

++  /**

++   * The break types it was constructed with

++   * @internal

++   */

++  uint32_t      fTypes;

++

++  /**

 +   * A Unicode set of all viramas

 +   * @internal

-    */

--  DictionaryBreakEngine();

++   */

 +  UnicodeSet    fViramaSet;

 +

 +  /**

@@ -751,10 +737,26 @@ diff -ur icu.org/source/common/dictbe.h icu/source/common/dictbe.h

 +  bool scanAfterEnd(UText *text, int32_t rangeEnd, int32_t& end, bool &doBreak) const;

 +  void scanBackClusters(UText *text, int32_t textStart, int32_t& start) const;

 +  void scanFwdClusters(UText *text, int32_t textEnd, int32_t& end) const;

- 

++

   public:

-@@ -83,7 +144,7 @@

+   /**

+-   * 
Constructor 

++   * 
Default constructor.

++   *

+    */

+   DictionaryBreakEngine();

+ 

+   /**

++   * 
Constructor with break types.

++   */

++  explicit DictionaryBreakEngine(uint32_t breakTypes);

++

++  /**

+    * 
Virtual destructor.

+    */

+   virtual ~DictionaryBreakEngine();

+@@ -68,7 +139,7 @@

     * 
Find any breaks within a run in the supplied text.

     *

     * @param text A UText representing the text. The iterator is left at

@@ -763,7 +765,7 @@ diff -ur icu.org/source/common/dictbe.h icu/source/common/dictbe.h

     * that starts from the first character in the range.

     * @param startPos The start of the run within the supplied text.

     * @param endPos The end of the run within the supplied text.

-@@ -245,118 +306,120 @@

+@@ -218,118 +289,120 @@

  };

@@ -997,8 +999,8 @@ diff -ur icu.org/source/common/dictbe.h icu/source/common/dictbe.h

  /*******************************************************************

 diff -ur icu.org/source/common/dictionarydata.cpp icu/source/common/dictionarydata.cpp

---- icu.org/source/common/dictionarydata.cpp	2017-01-20 01:20:31.000000000 +0100

-+++ icu/source/common/dictionarydata.cpp	2017-04-21 23:14:23.846894372 +0200

+--- icu.org/source/common/dictionarydata.cpp	2018-03-26 15:38:30.000000000 +0200

++++ icu/source/common/dictionarydata.cpp	2018-04-17 14:04:50.775567214 +0200

 @@ -44,7 +44,7 @@

  int32_t UCharsDictionaryMatcher::matches(UText *text, int32_t maxLength, int32_t limit,

@@ -1046,8 +1048,8 @@ diff -ur icu.org/source/common/dictionarydata.cpp icu/source/common/dictionaryda

                  if (values != NULL) {

                      values[wordCount] = bt.getValue();

 diff -ur icu.org/source/common/dictionarydata.h icu/source/common/dictionarydata.h

---- icu.org/source/common/dictionarydata.h	2017-01-20 01:20:31.000000000 +0100

-+++ icu/source/common/dictionarydata.h	2017-04-21 23:14:23.846894372 +0200

+--- icu.org/source/common/dictionarydata.h	2018-03-26 15:38:30.000000000 +0200

++++ icu/source/common/dictionarydata.h	2018-04-17 14:04:50.775567214 +0200

 @@ -21,6 +21,7 @@

  #include "unicode/utext.h"

  #include "unicode/udata.h"

@@ -1084,8 +1086,8 @@ diff -ur icu.org/source/common/dictionarydata.h icu/source/common/dictionarydata

  private:

      UChar32 transform(UChar32 c) const;

 diff -ur icu.org/source/data/Makefile.in icu/source/data/Makefile.in

---- icu.org/source/data/Makefile.in	2017-04-21 23:13:03.248087545 +0200

-+++ icu/source/data/Makefile.in	2017-04-21 23:14:23.846894372 +0200

+--- icu.org/source/data/Makefile.in	2018-04-17 12:28:37.098707466 +0200

++++ icu/source/data/Makefile.in	2018-04-17 14:04:50.775567214 +0200

 @@ -183,7 +183,7 @@

  endif

  endif

diff --git a/external/icu/icu4c-ubsan.patch.1 b/external/icu/icu4c-ubsan.patch.1

deleted file mode 100644

index 9f6aa3fbc9f8..000000000000

--- a/external/icu/icu4c-ubsan.patch.1

+++ /dev/null

@@ -1,91 +0,0 @@

-diff -ur icu.org/source/common/rbbidata.h icu/source/common/rbbidata.h

---- icu.org/source/common/rbbidata.h	2017-02-03 19:57:23.000000000 +0100

-+++ icu/source/common/rbbidata.h	2017-04-21 22:46:25.371651160 +0200

-@@ -115,7 +115,7 @@

-                                     /*     StatusTable of the set of matching             */

-                                     /*     tags (rule status values)                      */

-     int16_t          fReserved;

--    uint16_t         fNextState[2]; /*  Next State, indexed by char category.             */

-+    uint16_t         fNextState[1]; /*  Next State, indexed by char category.             */

-                                     /*  This array does not have two elements             */

-                                     /*    Array Size is actually fData->fHeader->fCatCount         */

-                                     /*    CAUTION:  see RBBITableBuilder::getTableSize()  */

-@@ -128,7 +128,7 @@

-     uint32_t         fRowLen;       /*  Length of a state table row, in bytes.            */

-     uint32_t         fFlags;        /*  Option Flags for this state table                 */

-     uint32_t         fReserved;     /*  reserved                                          */

--    char             fTableData[4]; /*  First RBBIStateTableRow begins here.              */

-+    char             fTableData[1]; /*  First RBBIStateTableRow begins here.              */

-                                     /*    (making it char[] simplifies ugly address       */

-                                     /*     arithmetic for indexing variable length rows.) */

- };

-diff -ur icu.org/source/common/rbbitblb.cpp icu/source/common/rbbitblb.cpp

---- icu.org/source/common/rbbitblb.cpp	2017-01-20 01:20:31.000000000 +0100

-+++ icu/source/common/rbbitblb.cpp	2017-04-21 22:46:25.373651159 +0200

-@@ -1095,15 +1095,15 @@

-         return 0;

-     }

- 

--    size    = sizeof(RBBIStateTable) - 4;    // The header, with no rows to the table.

-+    size    = offsetof(RBBIStateTable, fTableData);    // The header, with no rows to the table.

- 

-     numRows = fDStates->size();

-     numCols = fRB->fSetBuilder->getNumCharCategories();

- 

--    //  Note  The declaration of RBBIStateTableRow is for a table of two columns.

--    //        Therefore we subtract two from numCols when determining

-+    //  Note  The declaration of RBBIStateTableRow is for a table of one columns.

-+    //        Therefore we subtract one from numCols when determining

-     //        how much storage to add to a row for the total columns.

--    rowSize = sizeof(RBBIStateTableRow) + sizeof(uint16_t)*(numCols-2);

-+    rowSize = sizeof(RBBIStateTableRow) + sizeof(uint16_t)*(numCols-1);

-     size   += numRows * rowSize;

-     return size;

- }

-@@ -1133,7 +1133,7 @@

-     }

- 

-     table->fRowLen    = sizeof(RBBIStateTableRow) +

--                            sizeof(uint16_t) * (fRB->fSetBuilder->getNumCharCategories() - 2);

-+                            sizeof(uint16_t) * (fRB->fSetBuilder->getNumCharCategories() - 1);

-     table->fNumStates = fDStates->size();

-     table->fFlags     = 0;

-     if (fRB->fLookAheadHardBreak) {

-diff -ur icu.org/source/common/ubidiimp.h icu/source/common/ubidiimp.h

---- icu.org/source/common/ubidiimp.h	2017-02-03 19:57:23.000000000 +0100

-+++ icu/source/common/ubidiimp.h	2017-04-21 22:46:25.374651159 +0200

-@@ -198,8 +198,8 @@

- /* in a Run, logicalStart will get this bit set if the run level is odd */

- #define INDEX_ODD_BIT (1UL<<31)

- 

--#define MAKE_INDEX_ODD_PAIR(index, level) ((index)|((int32_t)(level)<<31))

--#define ADD_ODD_BIT_FROM_LEVEL(x, level)  ((x)|=((int32_t)(level)<<31))

-+#define MAKE_INDEX_ODD_PAIR(index, level) ((index)|((uint32_t)(level)<<31))

-+#define ADD_ODD_BIT_FROM_LEVEL(x, level)  ((x)|=((uint32_t)(level)<<31))

- #define REMOVE_ODD_BIT(x)                 ((x)&=~INDEX_ODD_BIT)

- 

- #define GET_INDEX(x)   ((x)&~INDEX_ODD_BIT)

-diff -ur icu.org/source/common/ucmndata.cpp icu/source/common/ucmndata.cpp

---- icu.org/source/common/ucmndata.cpp	2017-03-08 16:34:47.000000000 +0100

-+++ icu/source/common/ucmndata.cpp	2017-04-21 22:46:25.376651159 +0200

-@@ -77,7 +77,7 @@

- typedef struct  {

-     uint32_t          count;

-     uint32_t          reserved;

--    PointerTOCEntry   entry[2];   /* Actual size is from count. */

-+    PointerTOCEntry   entry[1];   /* Actual size is from count. */

- }  PointerTOC;

- 

- 

-diff -ur icu.org/source/common/ucmndata.h icu/source/common/ucmndata.h

---- icu.org/source/common/ucmndata.h	2017-01-20 01:20:31.000000000 +0100

-+++ icu/source/common/ucmndata.h	2017-04-21 22:46:25.377651159 +0200

-@@ -52,7 +52,7 @@

- 

- typedef struct {

-     uint32_t count;

--    UDataOffsetTOCEntry entry[2];    /* Actual size of array is from count. */

-+    UDataOffsetTOCEntry entry[1];    /* Actual size of array is from count. */

- } UDataOffsetTOC;

- 

- /**

--------------erAck-patch-parts--