ICU-2351 allow remove(guaranteed everything) and truncate(0) to also un-bogus a string

X-SVN-Rev: 10289

icu4c/source/common/caniter.cpp

@@ -148,11 +148,7 @@ UnicodeString CanonicalIterator::next() {
     }
 
     // delete old contents
-    if(buffer.isBogus()) {
-        buffer.setTo((UChar32)-1); // un-bogus the buffer
-    } else {
-        buffer.truncate(0); // keep the internal array allocated
-    }
+    buffer.remove();
 
     // construct return value
 

icu4c/source/common/unicode/unistr.h

@@ -1516,7 +1516,7 @@ public:
    * @draft ICU 2.0
    */
   int32_t
-  countChar32(int32_t start=0, int32_t length=0x7fffffff) const;
+  countChar32(int32_t start=0, int32_t length=INT32_MAX) const;
 
   /**
    * Check if the length UChar code units of the string
@@ -1531,7 +1531,7 @@ public:
    *
    * @param start the index of the first code unit to check (0 for the entire string)
    * @param length the number of UChar code units to check
-   *               (use 0x7fffffff for the entire string; remember that start/length
+   *               (use INT32_MAX for the entire string; remember that start/length
    *                values are pinned)
    * @param number The number of code points in the (sub)string is compared against
    *               the 'number' parameter.
@@ -1770,19 +1770,24 @@ public:
    * take a UErrorCode for simplicity.
    *
    * More generally, a bogus string has a non-value that is different from an empty string.
-   * It behaves much like a null reference in Java, a NULL pointer in C,
+   * It behaves much (but of course not entirely) like a null reference in Java, a NULL pointer in C,
    * or a NULL value in SQL.
    * A "bogus" string does not contain a string value, and getBuffer()
    * and similar will return NULL.
    *
    * The string object can be "revived" by assigning (operator=() or fastCopyFrom())
-   * another string, or by using one of the other setToXYZ functions.
+   * another string, or by using one of the other setToXYZ functions,
+   * or by using truncate() and remove() in ways that are always equivalent to assigning
+   * an empty string. See examples below.
    *
    * The simplest ways to turn a bogus string into an empty one
-   * is to assign an empty string, or to setTo an empty one.
-   * For example:
+   * is to use the remove() function.
+   * Examples for other functions that are equivalent to "set to empty string":
    * \code
    * if(s.isBogus()) {
+   *   s.remove();           // set to an empty string (remove all), or
+   *   s.remove(0, INT32_MAX); // set to an empty string (remove all), or
+   *   s.truncate(0);        // set to an empty string (complete truncation), or
    *   s=UnicodeString();    // assign an empty string, or
    *   s.setTo((UChar32)-1); // set to a pseudo code point that is out of range, or
    *   static const UChar nul=0;
@@ -3954,11 +3959,26 @@ UnicodeString::insert(int32_t start,
 inline UnicodeString& 
 UnicodeString::remove(int32_t start, 
              int32_t length)
-{ return doReplace(start, length, NULL, 0, 0); }
+{
+  if(start <= 0 && length == INT32_MAX) {
+    // remove(guaranteed everything) of a bogus string makes the string empty and non-bogus
+    return remove();
+  } else {
+    return doReplace(start, length, NULL, 0, 0);
+  }
+}
 
 inline UnicodeString& 
 UnicodeString::remove()
-{ return doReplace(0, fLength, 0, 0, 0); }
+{
+  // remove() of a bogus string makes the string empty and non-bogus
+  if(isBogus()) {
+    unBogus();
+  } else {
+    fLength = 0;
+  }
+  return *this;
+}
 
 inline UnicodeString& 
 UnicodeString::removeBetween(int32_t start,
@@ -3968,7 +3988,11 @@ UnicodeString::removeBetween(int32_t start,
 inline UBool 
 UnicodeString::truncate(int32_t targetLength)
 {
-  if((uint32_t)targetLength < (uint32_t)fLength) {
+  if(isBogus() && targetLength == 0) {
+    // truncate(0) of a bogus string makes the string empty and non-bogus
+    unBogus();
+    return FALSE;
+  } else if((uint32_t)targetLength < (uint32_t)fLength) {
     fLength = targetLength;
     return TRUE;
   } else {

icu4c/source/test/intltest/ustrtest.cpp

@@ -1257,32 +1257,38 @@ UnicodeStringTest::TestBogus() {
     }
 
     // verify that non-assignment modifications fail and do not revive a bogus string
+    test3.setToBogus();
     test3.append((UChar)0x61);
     if(!test3.isBogus() || test3.getBuffer()!=0) {
         errln("bogus.append('a') worked but must not");
     }
 
+    test3.setToBogus();
     test3.findAndReplace(UnicodeString((UChar)0x61), test2);
     if(!test3.isBogus() || test3.getBuffer()!=0) {
         errln("bogus.findAndReplace() worked but must not");
     }
 
+    test3.setToBogus();
     test3.trim();
     if(!test3.isBogus() || test3.getBuffer()!=0) {
         errln("bogus.trim() revived bogus but must not");
     }
 
-    test3.remove();
+    test3.setToBogus();
+    test3.remove(1);
     if(!test3.isBogus() || test3.getBuffer()!=0) {
-        errln("bogus.remove() revived bogus but must not");
+        errln("bogus.remove(1) revived bogus but must not");
     }
 
+    test3.setToBogus();
     if(!test3.setCharAt(0, 0x62).isBogus() || !test3.isEmpty()) {
         errln("bogus.setCharAt(0, 'b') worked but must not");
     }
 
-    if(test3.truncate(0) || !test3.isBogus() || !test3.isEmpty()) {
-        errln("bogus.truncate(0) revived bogus but must not");
+    test3.setToBogus();
+    if(test3.truncate(1) || !test3.isBogus() || !test3.isEmpty()) {
+        errln("bogus.truncate(1) revived bogus but must not");
     }
 
     // verify that assignments revive a bogus string
@@ -1337,7 +1343,7 @@ UnicodeStringTest::TestBogus() {
         errln("bogus.setTo(writable alias) failed");
     }
 
-    // same with simple, documented ways to turn a bogus string into an empty one
+    // verify simple, documented ways to turn a bogus string into an empty one
     test3.setToBogus();
     if(!test3.isBogus() || (test3=UnicodeString()).isBogus() || !test3.isEmpty()) {
         errln("bogus.operator=(UnicodeString()) failed");
@@ -1348,6 +1354,21 @@ UnicodeStringTest::TestBogus() {
         errln("bogus.setTo(UnicodeString()) failed");
     }
 
+    test3.setToBogus();
+    if(test3.remove().isBogus() || test3.getBuffer()==0 || !test3.isEmpty()) {
+        errln("bogus.remove() failed");
+    }
+
+    test3.setToBogus();
+    if(test3.remove(0, INT32_MAX).isBogus() || test3.getBuffer()==0 || !test3.isEmpty()) {
+        errln("bogus.remove(0, INT32_MAX) failed");
+    }
+
+    test3.setToBogus();
+    if(test3.truncate(0) || test3.isBogus() || !test3.isEmpty()) {
+        errln("bogus.truncate(0) failed");
+    }
+
     test3.setToBogus();
     if(!test3.isBogus() || test3.setTo((UChar32)-1).isBogus() || !test3.isEmpty()) {
         errln("bogus.setTo((UChar32)-1) failed");
@@ -1365,6 +1386,7 @@ UnicodeStringTest::TestBogus() {
         errln("setToBogus() failed to make a string bogus");
     }
 
+    test3.setToBogus();
     if(test1.isBogus() || !(test1=test3).isBogus()) {
         errln("normal=bogus failed to make the left string bogus");
     }