Update ICU to 58.1

source/common/locid.cpp

Index: source/common/locid.cpp
diff --git a/source/common/locid.cpp b/source/common/locid.cpp
index 27e403065f2e933c99c215e6d2de93f856c5981f..d2781db95bdd8cf8b4e0c31f9659406add9e299e 100644
--- a/source/common/locid.cpp
+++ b/source/common/locid.cpp
@@ -1,6 +1,8 @@
+// Copyright (C) 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
 /*
  **********************************************************************
- *   Copyright (C) 1997-2015, International Business Machines
+ *   Copyright (C) 1997-2016, International Business Machines
  *   Corporation and others.  All Rights Reserved.
  **********************************************************************
 *
@@ -42,6 +44,7 @@
 #include "uhash.h"
 #include "ucln_cmn.h"
 #include "ustr_imp.h"
+#include "charstr.h"
 
 U_CDECL_BEGIN
 static UBool U_CALLCONV locale_cleanup(void);
@@ -57,6 +60,12 @@ static UMutex gDefaultLocaleMutex = U_MUTEX_INITIALIZER;
 static UHashtable *gDefaultLocalesHashT = NULL;
 static Locale *gDefaultLocale = NULL;
 
+/**
+ * \def ULOC_STRING_LIMIT
+ * strings beyond this value crash in CharString
+ */
+#define ULOC_STRING_LIMIT 357913941
+
 U_NAMESPACE_END
 
 typedef enum ELocalePos {
@@ -283,13 +292,12 @@ Locale::Locale( const   char * newLanguage,
     }
     else
     {
-        MaybeStackArray<char, ULOC_FULLNAME_CAPACITY> togo;
+        UErrorCode status = U_ZERO_ERROR;
         int32_t size = 0;
         int32_t lsize = 0;
         int32_t csize = 0;
         int32_t vsize = 0;
         int32_t ksize = 0;
-        char    *p;
 
         // Calculate the size of the resulting string.
 
@@ -297,13 +305,23 @@ Locale::Locale( const   char * newLanguage,
         if ( newLanguage != NULL )
         {
             lsize = (int32_t)uprv_strlen(newLanguage);
+            if ( lsize < 0 || lsize > ULOC_STRING_LIMIT ) { // int32 wrap
+                setToBogus();
+                return;
+            }
             size = lsize;
         }
 
+        CharString togo(newLanguage, lsize, status); // start with newLanguage
+
         // _Country
         if ( newCountry != NULL )
         {
             csize = (int32_t)uprv_strlen(newCountry);
+            if ( csize < 0 || csize > ULOC_STRING_LIMIT ) { // int32 wrap
+                setToBogus();
+                return;
+            }
             size += csize;
         }
 
@@ -318,6 +336,10 @@ Locale::Locale( const   char * newLanguage,
 
             // remove trailing _'s
             vsize = (int32_t)uprv_strlen(newVariant);
+            if ( vsize < 0 || vsize > ULOC_STRING_LIMIT ) { // int32 wrap
+                setToBogus();
+                return;
+            }
             while( (vsize>1) && (newVariant[vsize-1] == SEP_CHAR) )
             {
                 vsize--;
@@ -342,70 +364,56 @@ Locale::Locale( const   char * newLanguage,
         if ( newKeywords != NULL)
         {
             ksize = (int32_t)uprv_strlen(newKeywords);
+            if ( ksize < 0 || ksize > ULOC_STRING_LIMIT ) {
+              setToBogus();
+              return;
+            }
             size += ksize + 1;
         }
 
-
         //  NOW we have the full locale string..
-
-        /*if the whole string is longer than our internal limit, we need
-        to go to the heap for temporary buffers*/
-        if (size >= togo.getCapacity())
-        {
-            // If togo_heap could not be created, initialize with default settings.
-            if (togo.resize(size+1) == NULL) {
-                init(NULL, FALSE);
-            }
-        }
-
-        togo[0] = 0;
-
         // Now, copy it back.
-        p = togo.getAlias();
-        if ( lsize != 0 )
-        {
-            uprv_strcpy(p, newLanguage);
-            p += lsize;
-        }
+
+        // newLanguage is already copied
 
         if ( ( vsize != 0 ) || (csize != 0) )  // at least:  __v
         {                                      //            ^
-            *p++ = SEP_CHAR;
+            togo.append(SEP_CHAR, status);
         }
 
         if ( csize != 0 )
         {
-            uprv_strcpy(p, newCountry);
-            p += csize;
+            togo.append(newCountry, status);
         }
 
         if ( vsize != 0)
         {
-            *p++ = SEP_CHAR; // at least: __v
-
-            uprv_strncpy(p, newVariant, vsize);  // Must use strncpy because
-            p += vsize;                          // of trimming (above).
-            *p = 0; // terminate
+            togo.append(SEP_CHAR, status)
+                .append(newVariant, vsize, status);
         }
 
         if ( ksize != 0)
         {
             if (uprv_strchr(newKeywords, '=')) {
-                *p++ = '@'; /* keyword parsing */
+                togo.append('@', status); /* keyword parsing */
             }
             else {
-                *p++ = '_'; /* Variant parsing with a script */
+                togo.append('_', status); /* Variant parsing with a script */
                 if ( vsize == 0) {
-                    *p++ = '_'; /* No country found */
+                    togo.append('_', status); /* No country found */
                 }
             }
-            uprv_strcpy(p, newKeywords);
-            p += ksize;
+            togo.append(newKeywords, status);
         }
 
+        if (U_FAILURE(status)) {
+            // Something went wrong with appending, etc.
+            setToBogus();
+            return;
+        }
         // Parse it, because for example 'language' might really be a complete
         // string.
-        init(togo.getAlias(), FALSE);
+        init(togo.data(), FALSE);
     }
 }
 
@@ -536,7 +544,7 @@ Locale& Locale::init(const char* localeID, UBool canonicalize)
         /* after uloc_getName/canonicalize() we know that only '_' are separators */
         separator = field[0] = fullName;
         fieldIdx = 1;
-        while ((separator = uprv_strchr(field[fieldIdx-1], SEP_CHAR)) && fieldIdx < (int32_t)(sizeof(field)/sizeof(field[0]))-1) {
+        while ((separator = uprv_strchr(field[fieldIdx-1], SEP_CHAR)) && fieldIdx < UPRV_LENGTHOF(field)-1) {
             field[fieldIdx] = separator + 1;
             fieldLen[fieldIdx-1] = (int32_t)(separator - field[fieldIdx-1]);
             fieldIdx++;
@@ -662,6 +670,7 @@ Locale::setToBogus() {
     *script = 0;
     *country = 0;
     fIsBogus = TRUE;
+    variantBegin = 0;
 }
 
 const Locale& U_EXPORT2