ICU-491.11.1.tar.gz

[apple/icu.git] / icuSources / samples / ucnv / convsamp.cpp

diff --git a/icuSources/samples/ucnv/convsamp.cpp b/icuSources/samples/ucnv/convsamp.cpp
index 506a1b6448b65e48a76044d32149a04232cf9d5f..63922c773639c613f8b916adc23a9718316ce50a 100644 (file)
--- a/icuSources/samples/ucnv/convsamp.cpp
+++ b/icuSources/samples/ucnv/convsamp.cpp
@@ -1,6 +1,6 @@
 /**************************************************************************
 *
-*   Copyright (C) 2000-2010, International Business Machines
+*   Copyright (C) 2000-2011, International Business Machines
 *   Corporation and others.  All Rights Reserved.
 *
 ***************************************************************************
@@ -58,20 +58,13 @@ void prettyPrintUChar(UChar c)
     char buf[1000];
     UErrorCode status = U_ZERO_ERROR;
     int32_t o;
-    
-    o = u_charName(c, U_UNICODE_CHAR_NAME, buf, 1000, &status);
+
+    o = u_charName(c, U_EXTENDED_CHAR_NAME, buf, 1000, &status);
     if(U_SUCCESS(status) && (o>0) ) {
       buf[6] = 0;
       printf("%7s", buf);
     } else {
-      o = u_charName(c, U_UNICODE_10_CHAR_NAME, buf, 1000, &status);
-      if(U_SUCCESS(status) && (o>0)) {
-        buf[5] = 0;
-        printf("~%6s", buf);
-      }
-      else {
-        printf(" ??????");
-      }
+      printf(" ??????");
     }
   } else {
     switch((char)(c & 0x007F)) {
@@ -212,7 +205,9 @@ UErrorCode convsample_02()
   int32_t     len;
 
   // set up the converter
+  //! [ucnv_open]
   conv = ucnv_open("koi8-r", &status);
+  //! [ucnv_open]
   assert(U_SUCCESS(status));
 
   // convert to koi8-r
@@ -397,7 +392,6 @@ UErrorCode convsample_06()
   char inBuf[BUFFERSIZE];
   const char *source;
   const char *sourceLimit;
-  UChar *uBuf;
   int32_t uBufSize = 0;
   UConverter *conv;
   UErrorCode status = U_ZERO_ERROR;
@@ -438,8 +432,6 @@ UErrorCode convsample_06()
   uBufSize = (BUFFERSIZE/ucnv_getMinCharSize(conv));
   printf("input bytes %d / min chars %d = %d UChars\n",
          BUFFERSIZE, ucnv_getMinCharSize(conv), uBufSize);
-  uBuf = (UChar*)malloc(uBufSize * sizeof(UChar));
-  assert(uBuf!=NULL);
 
   // grab another buffer's worth
   while((!feof(f)) && 
@@ -473,6 +465,9 @@ UErrorCode convsample_06()
       if(p>charCount)
       {
         fprintf(stderr, "U+%06X: oh.., we only handle BMP characters so far.. redesign!\n", p);
+        free(info);
+        fclose(f);
+        ucnv_close(conv);
         return U_UNSUPPORTED_ERROR;
       }
       info[p].frequency++;
@@ -1073,6 +1068,42 @@ UErrorCode convsample_46()
 
 #define BUFFERSIZE 219
 
+void convsample_50() {
+  printf("\n\n==============================================\n"
+         "Sample 50: C: ucnv_detectUnicodeSignature\n");
+
+  //! [ucnv_detectUnicodeSignature]
+  UErrorCode err = U_ZERO_ERROR;
+  UBool discardSignature = TRUE; /* set to TRUE to throw away the initial U+FEFF */
+  char input[] = { '\xEF','\xBB', '\xBF','\x41','\x42','\x43' };
+  int32_t signatureLength = 0;
+  const char *encoding = ucnv_detectUnicodeSignature(input,sizeof(input),&signatureLength,&err);
+  UConverter *conv = NULL;
+  UChar output[100];
+  UChar *target = output, *out;
+  const char *source = input;
+  if(encoding!=NULL && U_SUCCESS(err)){
+    // should signature be discarded ?
+    conv = ucnv_open(encoding, &err);
+    // do the conversion
+    ucnv_toUnicode(conv,
+                   &target, output + sizeof(output)/U_SIZEOF_UCHAR,
+                   &source, input + sizeof(input),
+                   NULL, TRUE, &err);
+    out = output;
+    if (discardSignature){
+      ++out; // ignore initial U+FEFF
+    }
+    while(out != target) {
+      printf("%04x ", *out++);
+    }
+    puts("");
+  }
+  //! [ucnv_detectUnicodeSignature]
+  puts("");
+}
+
+
 
 /* main */
 
@@ -1096,6 +1127,8 @@ int main()
   convsample_40();  // C,   cp37 -> UTF16 [data02.bin -> data40.utf16]
   
   convsample_46();  // C,  UTF16 -> latin3 [data41.utf16 -> data46.out]
+
+  convsample_50();  // C, detect unicode signature
   
   printf("End of converter samples.\n");