Index: /icu/branches/maint/maint-3-8/source/i18n/regexcmp.cpp

===================================================================

--- i18n/regexcmp.cpp (revision 21805)

+++ i18n/regexcmp.cpp (revision 23292)

@@ -3,5 +3,5 @@

 //  file:  regexcmp.cpp

 //

-//  Copyright (C) 2002-2007 International Business Machines Corporation and others.

+//  Copyright (C) 2002-2008 International Business Machines Corporation and others.

 //  All Rights Reserved.

 //

@@ -1187,12 +1187,15 @@

             //  we fill the operand with the capture group number.  At the end

             //  of compilation, it will be changed to the variable's location.

-            U_ASSERT(groupNum > 0);

-            int32_t  op;

-            if (fModeFlags & UREGEX_CASE_INSENSITIVE) {

-                op = URX_BUILD(URX_BACKREF_I, groupNum);

+            if (groupNum < 1) { 

+                error(U_REGEX_INVALID_BACK_REF);

             } else {

-                op = URX_BUILD(URX_BACKREF, groupNum);

-            }

-            fRXPat->fCompiledPat->addElement(op, *fStatus);

+                int32_t  op;

+                if (fModeFlags & UREGEX_CASE_INSENSITIVE) {

+                    op = URX_BUILD(URX_BACKREF_I, groupNum);

+                } else {

+                    op = URX_BUILD(URX_BACKREF, groupNum);

+                }

+                fRXPat->fCompiledPat->addElement(op, *fStatus);

+            }

         }

         break;

Index: /icu/branches/maint/maint-3-8/source/i18n/rematch.cpp

===================================================================

--- i18n/rematch.cpp (revision 21973)

+++ i18n/rematch.cpp (revision 23292)

@@ -1,5 +1,5 @@

 /*

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

-*   Copyright (C) 2002-2007 International Business Machines Corporation  *

+*   Copyright (C) 2002-2008 International Business Machines Corporation  *

 *   and others. All rights reserved.                                     *

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

@@ -30,4 +30,13 @@

 U_NAMESPACE_BEGIN

+

+// Limit the size of the back track stack, to avoid system failures caused

+//   by heap exhaustion.  Units are in 32 bit words, not bytes.

+// This value puts ICU's limits higher than most other regexp implementations,

+//  which use recursion rather than the heap, and take more storage per

+//  backtrack point.

+// This constant is _temporary_.  Proper API to control the value will added.

+//

+static const int32_t BACKTRACK_STACK_CAPACITY = 8000000;

 //-----------------------------------------------------------------------------

@@ -54,6 +63,7 @@

     if (fStack == NULL || fData == NULL) {

         fDeferredStatus = U_MEMORY_ALLOCATION_ERROR;

-    }

-        

+    } else {

+        fStack->setMaxCapacity(BACKTRACK_STACK_CAPACITY);

+    }

     reset(RegexStaticSets::gStaticSets->fEmptyString);

 }

@@ -79,4 +89,6 @@

     if (fStack == NULL || fData == NULL) {

         status = U_MEMORY_ALLOCATION_ERROR;

+    } else {

+        fStack->setMaxCapacity(BACKTRACK_STACK_CAPACITY);

     }

     reset(input);

@@ -103,4 +115,6 @@

     if (fStack == NULL || fData == NULL) {

         status = U_MEMORY_ALLOCATION_ERROR;

+    } else {

+        fStack->setMaxCapacity(BACKTRACK_STACK_CAPACITY);

     }

     reset(RegexStaticSets::gStaticSets->fEmptyString);

@@ -1015,4 +1029,12 @@

     // push storage for a new frame. 

     int32_t *newFP = fStack->reserveBlock(frameSize, status);

+    if (newFP == NULL) {

+        // Heap allocation error on attempted stack expansion.

+        // We need to return a writable stack frame, so just return the

+        //    previous frame.  The match operation will stop quickly

+        //    becuase of the error status, after which the frame will never

+        //    be looked at again.

+        return fp;

+    }

     fp = (REStackFrame *)(newFP - frameSize);  // in case of realloc of stack.

@@ -1030,6 +1052,6 @@

     return (REStackFrame *)newFP;

 }

-    

-            

+

+

 //--------------------------------------------------------------------------------

 //

@@ -2262,4 +2284,5 @@

         if (U_FAILURE(status)) {

+            isMatch = FALSE;

             break;

         }

Index: /icu/branches/maint/maint-3-8/source/test/intltest/regextst.h

===================================================================

--- test/intltest/regextst.h (revision 22001)

+++ test/intltest/regextst.h (revision 23292)

@@ -1,5 +1,5 @@

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

  * COPYRIGHT: 

- * Copyright (c) 2002-2007, International Business Machines Corporation and

+ * Copyright (c) 2002-2008, International Business Machines Corporation and

  * others. All Rights Reserved.

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

@@ -31,4 +31,5 @@

     virtual void Errors();

     virtual void PerlTests();

+    virtual void Bug6149();

     // The following functions are internal to the regexp tests.

Index: /icu/branches/maint/maint-3-8/source/test/intltest/regextst.cpp

===================================================================

--- test/intltest/regextst.cpp (revision 22057)

+++ test/intltest/regextst.cpp (revision 23292)

@@ -1,5 +1,5 @@

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

  * COPYRIGHT:

- * Copyright (c) 2002-2007, International Business Machines Corporation and

+ * Copyright (c) 2002-2008, International Business Machines Corporation and

  * others. All Rights Reserved.

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

@@ -67,4 +67,8 @@

             if (exec) PerlTests();

             break;

+        case 7: name = "Bug 6149";

+            if (exec) Bug6149();

+            break;

+            

@@ -1640,4 +1644,10 @@

     // Ticket 5389

     REGEX_ERR("*c", 1, 1, U_REGEX_RULE_SYNTAX);

+    

+    // Invalid Back Reference \0

+    //    For ICU 3.8 and earlier

+    //    For ICU versions newer than 3.8, \0 introduces an octal escape.

+    //

+    REGEX_ERR("(ab)\\0", 1, 6, U_REGEX_INVALID_BACK_REF);

 }

@@ -2123,4 +2133,24 @@

+//--------------------------------------------------------------

+//

+//  Bug6149   Verify limits to heap expansion for backtrack stack.

+//             Use this pattern,

+//                 "(a?){1,}"

+//             The zero-length match will repeat forever.

+//                (That this goes into a loop is another bug)

+//

+//---------------------------------------------------------------

+void RegexTest::Bug6149() {

+    UnicodeString pattern("(a?){1,}");

+    UnicodeString s("xyz");

+    uint32_t flags = 0;

+    UErrorCode status = U_ZERO_ERROR;

+    

+    RegexMatcher  matcher(pattern, s, flags, status);

+    UBool result = false;

+    REGEX_ASSERT_FAIL(result=matcher.matches(status), U_BUFFER_OVERFLOW_ERROR);

+    REGEX_ASSERT(result == FALSE);

+ }

 #endif  /* !UCONFIG_NO_REGULAR_EXPRESSIONS  */

Index: /icu/branches/maint/maint-3-8/source/common/uvectr32.cpp

===================================================================

--- common/uvectr32.cpp (revision 12958)

+++ common/uvectr32.cpp (revision 23292)

@@ -1,5 +1,5 @@

 /*

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

-* Copyright (C) 1999-2003, International Business Machines Corporation and   *

+* Copyright (C) 1999-2008, International Business Machines Corporation and   *

 * others. All Rights Reserved.                                               *

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

@@ -27,4 +27,5 @@

     count(0),

     capacity(0),

+    maxCapacity(0),

     elements(NULL)

 {

@@ -35,4 +36,5 @@

     count(0),

     capacity(0),

+    maxCapacity(0),

     elements(0)

 {

@@ -46,4 +48,7 @@

     if (initialCapacity < 1) {

         initialCapacity = DEFUALT_CAPACITY;

+    }

+    if (maxCapacity>0 && maxCapacity<initialCapacity) {

+        initialCapacity = maxCapacity;

     }

     elements = (int32_t *)uprv_malloc(sizeof(int32_t)*initialCapacity);

@@ -190,19 +195,33 @@

     if (capacity >= minimumCapacity) {

         return TRUE;

-    } else {

-        int32_t newCap = capacity * 2;

-        if (newCap < minimumCapacity) {

-            newCap = minimumCapacity;

-        }

-        int32_t* newElems = (int32_t *)uprv_malloc(sizeof(int32_t)*newCap);

-        if (newElems == 0) {

-            status = U_MEMORY_ALLOCATION_ERROR;

-            return FALSE;

-        }

-        uprv_memcpy(newElems, elements, sizeof(elements[0]) * count);

-        uprv_free(elements);

-        elements = newElems;

-        capacity = newCap;

-        return TRUE;

+    }

+    if (maxCapacity>0 && minimumCapacity>maxCapacity) {

+        status = U_BUFFER_OVERFLOW_ERROR;

+        return FALSE;

+    }

+    int32_t newCap = capacity * 2;

+    if (newCap < minimumCapacity) {

+        newCap = minimumCapacity;

+    }

+    if (maxCapacity > 0 && newCap > maxCapacity) {

+        newCap = maxCapacity;

+    }

+    int32_t* newElems = (int32_t *)uprv_malloc(sizeof(int32_t)*newCap);

+    if (newElems == 0) {

+        status = U_MEMORY_ALLOCATION_ERROR;

+        return FALSE;

+    }

+    uprv_memcpy(newElems, elements, sizeof(elements[0]) * count);

+    uprv_free(elements);

+    elements = newElems;

+    capacity = newCap;

+    return TRUE;

+}

+

+void UVector32::setMaxCapacity(int32_t limit) {

+    U_ASSERT(limit >= 0);

+    maxCapacity = limit;

+    if (maxCapacity < 0) {

+        maxCapacity = 0;

     }

 }

Index: /icu/branches/maint/maint-3-8/source/common/uvectr32.h

===================================================================

--- common/uvectr32.h (revision 19000)

+++ common/uvectr32.h (revision 23292)

@@ -1,5 +1,5 @@

 /*

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

-*   Copyright (C) 1999-2006, International Business Machines

+*   Copyright (C) 1999-2008, International Business Machines

 *   Corporation and others.  All Rights Reserved.

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

@@ -62,4 +62,6 @@

     int32_t   capacity;

+    

+    int32_t   maxCapacity;   // Limit beyond which capacity is not permitted to grow.

     int32_t*  elements;

@@ -161,4 +163,12 @@

      */

     int32_t *getBuffer() const;

+

+    /**

+     * Set the maximum allowed buffer capacity for this vector/stack.

+     * Default with no limit set is unlimited, go until malloc() fails.

+     * A Limit of zero means unlimited capacity.

+     * Units are vector elements (32 bits each), not bytes.

+     */

+    void setMaxCapacity(int32_t limit);

     /**

@@ -222,5 +232,7 @@

 inline int32_t *UVector32::reserveBlock(int32_t size, UErrorCode &status) {

-    ensureCapacity(count+size, status);

+    if (ensureCapacity(count+size, status) == FALSE) {

+        return NULL;

+    }

     int32_t  *rp = elements+count;

     count += size;

--- runConfigureICU	Fri Jan 16 17:26:40 2004

+++ runConfigureICU	Mon Nov 15 11:05:50 2004

@@ -289,7 +289,7 @@

     *BSD)

-        THE_OS="BSD"

+        THE_OS="$platform"

-        THE_COMP="the GNU C++"

-        CC=gcc; export CC

-        CXX=g++; export CXX

+        CC=${CC:=cc}; export CC

+        CXX=${CXX:=c++}; export CXX

+        THE_COMP="`$CC -v | fgrep -i ver`" # works with icc and gcc at least

         ;;

     QNX)

@@ -358,12 +358,4 @@

 if test $release -eq 1

 then

-    if test "$RELEASE_CFLAGS" = ""

-    then

-        case $CC in

-            gcc|*/gcc|*-gcc-*|*/*-gcc-*)

-                RELEASE_CFLAGS=-O3

-                ;;

-        esac

-    fi

     if test "$RELEASE_CFLAGS" != ""

     then

Fix self-test(s) breaking, when build happens on days with other than

24-hours (such as when daylight saving time kicks in). See:

	http://www.freebsd.org/cgi/query-pr.cgi?pr=128521

	http://www.icu-project.org/trac/ticket/6018

Index: /icu/trunk/source/i18n/calendar.cpp

===================================================================

--- i18n/calendar.cpp (revision 22978)

+++ i18n/calendar.cpp (revision 23166)

@@ -1,5 +1,5 @@

 /*

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

-* Copyright (C) 1997-2007, International Business Machines Corporation and    *

+* Copyright (C) 1997-2008, International Business Machines Corporation and    *

 * others. All Rights Reserved.                                                *

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

@@ -74,6 +74,12 @@

 #if defined( U_DEBUG_CALSVC ) || defined (U_DEBUG_CAL)

+

+/** 

+ * fldName was removed as a duplicate implementation. 

+ * use  udbg_ services instead, 

+ * which depend on include files and library from ../tools/ctestfw

+ */

+#include "unicode/udbgutil.h"

 #include <stdio.h>

-

 /**

@@ -83,12 +89,8 @@

 * @internal

 */

-#error fldName() has been removed. Please use udbg_ucal_fieldName()  from libctestfw instead. The following code might work.

-

-static const char* fldName(UCalendarDateFields f) {

-	const char *udbg_ucal_fieldName(int32_t fld);

-	return udbg_ucal_fieldName((int32_t)f);

-}

-

-

+

+const char* fldName(UCalendarDateFields f) {

+	return udbg_enumName(UDBG_UCalendarDateFields, (int32_t)f);

+}

 #if UCAL_DEBUG_DUMP

Index: /icu/trunk/source/i18n/reldtfmt.cpp

===================================================================

--- i18n/reldtfmt.cpp (revision 22561)

+++ i18n/reldtfmt.cpp (revision 23166)

@@ -1,5 +1,5 @@

 /*

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

-* Copyright (C) 2007, International Business Machines Corporation and    *

+* Copyright (C) 2007-2008, International Business Machines Corporation and    *

 * others. All Rights Reserved.                                                *

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

@@ -304,8 +304,13 @@

         return 0;

     }

-    // TODO: Cache the nowCal to avoid heap allocs?

+    // TODO: Cache the nowCal to avoid heap allocs? Would be difficult, don't know the calendar type

     Calendar *nowCal = cal.clone();

     nowCal->setTime(Calendar::getNow(), status);

-    int32_t dayDiff = nowCal->fieldDifference(cal.getTime(status), Calendar::DATE, status);

+

+    // For the day difference, we are interested in the difference in the (modified) julian day number

+    // which is midnight to midnight.  Using fieldDifference() is NOT correct here, because 

+    // 6pm Jan 4th  to 10am Jan 5th should be considered "tomorrow".

+    int32_t dayDiff = cal.get(UCAL_JULIAN_DAY, status) - nowCal->get(UCAL_JULIAN_DAY, status);

+

     delete nowCal;

     return dayDiff;

Index: /icu/trunk/source/test/intltest/dadrfmt.cpp

===================================================================

--- test/intltest/dadrfmt.cpp (revision 22493)

+++ test/intltest/dadrfmt.cpp (revision 23166)

@@ -1,5 +1,5 @@

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

  * COPYRIGHT: 

- * Copyright (c) 1997-2007, International Business Machines Corporation and

+ * Copyright (c) 1997-2008, International Business Machines Corporation and

  * others. All Rights Reserved.

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

@@ -94,4 +94,5 @@

     UnicodeString kMILLIS("MILLIS="); // TODO: static

     UnicodeString kRELATIVE_MILLIS("RELATIVE_MILLIS="); // TODO: static

+    UnicodeString kRELATIVE_ADD("RELATIVE_ADD:"); // TODO: static

     UErrorCode status = U_ZERO_ERROR;

@@ -170,5 +171,10 @@

             }

         }

-        

+

+        Calendar *cal = Calendar::createInstance(loc, status);

+        if(U_FAILURE(status)) {

+            errln("case %d: could not create calendar from %s", n, calLoc);

+        }

+

         // parse 'date'

         if(date.startsWith(kMILLIS)) {

@@ -180,4 +186,28 @@

             useDate = TRUE;

             fromDate = udbg_stoi(millis) + now;

+        } else if(date.startsWith(kRELATIVE_ADD)) {

+            UnicodeString add = UnicodeString(date, kRELATIVE_ADD.length());  // "add" is a string indicating which fields to add

+            if(fromSet.parseFrom(add, status)<0 || U_FAILURE(status)) {

+                errln("case %d: could not parse date as RELATIVE_ADD calendar fields: %s", n, u_errorName(status));

+                continue;

+            }

+            logln("Parsing ..\n");

+            useDate=TRUE;

+            cal->clear();

+            cal->setTime(now, status);

+            for (int q=0; q<UCAL_FIELD_COUNT; q++) {

+                if (fromSet.isSet((UCalendarDateFields)q)) {

+                    int32_t oldv = cal->get((UCalendarDateFields)q, status);

+                    cal->add((UCalendarDateFields)q,

+                                fromSet.get((UCalendarDateFields)q), status);

+                    int32_t newv = cal->get((UCalendarDateFields)q, status);

+                    logln("adding %d to %s ..went from %d to %d\n", fromSet.get((UCalendarDateFields)q), udbg_enumName(UDBG_UCalendarDateFields, q), oldv, newv);

+                }

+            }

+            fromDate = cal->getTime(status);

+            if(U_FAILURE(status)) {

+                errln("case %d: could not apply date as RELATIVE_ADD calendar fields: %s", n, u_errorName(status));

+                continue;

+            }

         } else if(fromSet.parseFrom(date, status)<0 || U_FAILURE(status)) {

             errln("case %d: could not parse date as calendar fields: %s", n, u_errorName(status));

@@ -185,8 +215,4 @@

         }

-        Calendar *cal = Calendar::createInstance(loc, status);

-        if(U_FAILURE(status)) {

-            errln("case %d: could not create calendar from %s", n, calLoc);

-        }

         // now, do it.

         if (fmt) {

Index: /icu/trunk/source/test/testdata/format.txt

===================================================================

--- test/testdata/format.txt (revision 22885)

+++ test/testdata/format.txt (revision 23166)

@@ -1,3 +1,3 @@

-// Copyright (c) 2007 International Business Machines

+// Copyright (c) 2007-2008 International Business Machines

 // Corporation and others. All Rights Reserved.

 format:table(nofallback) {

@@ -23,5 +23,8 @@

             // locale: locale including calendar type

             // spec:   either 'PATTERN=y mm h' etc, or 'DATE=SHORT,TIME=LONG'

-            // date:   either 'MILLIS=####' where #### is an unsigned long (millis), or a calendar spec ERA=0,YEAR=1, etc.. applied to the calendar type specified by the locale

+            // date:   either 'MILLIS=####' where #### is millis,

+            //        or a calendar spec ERA=0,YEAR=1, etc.. applied to the calendar type specified by the locale

+            //        or RELATIVE_MILLIS=### where ### is a signed value which is added to the current millis

+            //        or RELATIVE_ADD:DATE=1  which means that the field "DATE" will be added by +1 relative to current time

             // str:   the expected unicode string

             Cases { 

@@ -61,5 +64,5 @@

                     "en_US@calendar=gregorian",         

                     "DATE=RELATIVE_SHORT",

-                    "RELATIVE_MILLIS=86410000", // one day from now

+                    "RELATIVE_ADD:DATE=1", // one day from now

                     "Tomorrow"

                },

@@ -67,5 +70,5 @@

                     "en_US@calendar=gregorian",         

                     "DATE=RELATIVE_SHORT",

-                    "RELATIVE_MILLIS=0", // one day before now

+                    "RELATIVE_MILLIS=0", // today

                     "Today"

                },

@@ -73,5 +76,5 @@

                     "en_US@calendar=gregorian",         

                     "DATE=RELATIVE_SHORT",

-                    "RELATIVE_MILLIS=-86410000", // one day before now

+                    "RELATIVE_ADD:DATE=-1", // one day before now

                     "Yesterday"

                },

@@ -79,5 +82,5 @@

                     "mt_MT@calendar=gregorian",         

                     "DATE=RELATIVE_SHORT",

-                    "RELATIVE_MILLIS=86410000", // one day from now

+                    "RELATIVE_ADD:DATE=1", // one day from now

                     "Għada"

                },

@@ -85,5 +88,5 @@

                     "mt_MT@calendar=gregorian",         

                     "DATE=RELATIVE_SHORT",

-                    "RELATIVE_MILLIS=0", // one day before now

+                    "RELATIVE_MILLIS=0", // today

                     "Illum"

                },

@@ -91,5 +94,5 @@

                     "mt_MT@calendar=gregorian",         

                     "DATE=RELATIVE_SHORT",

-                    "RELATIVE_MILLIS=-86410000", // one day before now

+                    "RELATIVE_ADD:DATE=-1", // one day before now

                     "Lbieraħ"

                },

@@ -97,5 +100,5 @@

                     "ru",

                     "DATE=RELATIVE_SHORT",

-                    "RELATIVE_MILLIS=-172810000", // 2 days ago

+                    "RELATIVE_ADD:DATE=-2", // 2 days ago

                     "Позавчера"

                },

--- Makefile.in	2007-12-12 13:58:06.000000000 -0500

+++ Makefile.in	2008-01-28 15:20:56.000000000 -0500

@@ -126,5 +126,4 @@

 	$(INSTALL_DATA) @platform_make_fragment@ $(DESTDIR)$(pkgdatadir)/config/@platform_make_fragment_name@

 	$(INSTALL_SCRIPT) $(top_srcdir)/mkinstalldirs $(DESTDIR)$(pkgdatadir)/mkinstalldirs

-	$(INSTALL_SCRIPT) $(top_srcdir)/install-sh $(DESTDIR)$(pkgdatadir)/install-sh

 	$(INSTALL_DATA) $(top_srcdir)/../license.html $(DESTDIR)$(pkgdatadir)/license.html

 	$(INSTALL_SCRIPT) $(top_builddir)/config/icu-config $(DESTDIR)$(bindir)/icu-config

--- common/putil.c	2007-12-12 13:57:26.000000000 -0500

+++ common/putil.c	2008-01-28 13:04:02.000000000 -0500

@@ -577,5 +577,5 @@

 /* These platforms are likely to use Olson timezone IDs. */

 #define CHECK_LOCALTIME_LINK 1

-#if defined(U_LINUX)

+#if defined(U_LINUX) || defined(U_BSD)

 #define TZDEFAULT       "/etc/localtime"

 #define TZZONEINFO      "/usr/share/zoneinfo/"

sbin/genccode

sbin/gencmn

sbin/gensprep

sbin/genuca

sbin/icupkg

sbin/icuswap

@dirrm include/unicode

%%DATADIR%%/%%ICUVER%%/mkinstalldirs