finished adding UTF16_MUNGED charset

Changed the approach for the charset to go via utf16, which makes a
bit more sense to read. 

Added a testsuiite for UTF16_MUNGED as part of LOCAL-ICONV

4 files changed, 102 insertions, 72 deletions

diff --git a/lib/util/charset/charcnv.c b/lib/util/charset/charcnv.c
index 2ae16c3250..9dd68f05ea 100644
--- a/lib/util/charset/charcnv.c
+++ b/lib/util/charset/charcnv.c
@@ -57,6 +57,7 @@ static const char *charset_name(struct smb_iconv_convenience *ic, charset_t ch)
 	case CH_DOS: return ic->dos_charset;
 	case CH_UTF8: return "UTF8";
 	case CH_UTF16BE: return "UTF-16BE";
+	case CH_UTF16MUNGED: return "UTF16_MUNGED";
 	default:
 	return "ASCII";
 	}
diff --git a/lib/util/charset/charset.h b/lib/util/charset/charset.h
index 21fc20b8c3..cace79f949 100644
--- a/lib/util/charset/charset.h
+++ b/lib/util/charset/charset.h
@@ -28,9 +28,9 @@
 #include <talloc.h>
 
 /* this defines the charset types used in samba */
-typedef enum {CH_UTF16=0, CH_UNIX, CH_DOS, CH_UTF8, CH_UTF16BE} charset_t;
+typedef enum {CH_UTF16=0, CH_UNIX, CH_DOS, CH_UTF8, CH_UTF16BE, CH_UTF16MUNGED} charset_t;
 
-#define NUM_CHARSETS 5
+#define NUM_CHARSETS 6
 
 /*
  *   for each charset we have a function that pulls from that charset to
diff --git a/lib/util/charset/iconv.c b/lib/util/charset/iconv.c
index 10b3a6488b..b6842a49aa 100644
--- a/lib/util/charset/iconv.c
+++ b/lib/util/charset/iconv.c
@@ -51,7 +51,7 @@ static size_t ascii_pull  (void *,const char **, size_t *, char **, size_t *);
 static size_t ascii_push  (void *,const char **, size_t *, char **, size_t *);
 static size_t utf8_pull   (void *,const char **, size_t *, char **, size_t *);
 static size_t utf8_push   (void *,const char **, size_t *, char **, size_t *);
-static size_t utf8_munged_push(void *,const char **, size_t *, char **, size_t *);
+static size_t utf16_munged_pull(void *,const char **, size_t *, char **, size_t *);
 static size_t ucs2hex_pull(void *,const char **, size_t *, char **, size_t *);
 static size_t ucs2hex_push(void *,const char **, size_t *, char **, size_t *);
 static size_t iconv_copy  (void *,const char **, size_t *, char **, size_t *);
@@ -69,7 +69,7 @@ static const struct charset_functions builtin_functions[] = {
 	{"UTF-8",   utf8_pull,  utf8_push},
 
 	/* this handles the munging needed for String2Key */
-	{"UTF8_MUNGED",   utf8_pull,  utf8_munged_push},
+	{"UTF16_MUNGED",   utf16_munged_pull,  iconv_copy},
 
 	{"ASCII", ascii_pull, ascii_push},
 	{"UCS2-HEX", ucs2hex_pull, ucs2hex_push}
@@ -713,103 +713,74 @@ error:
 
 
 /*
-  this takes a UTF16 sequence, munges it according to the string2key
-  rules, and produces a UTF8 sequence
+  this takes a UTF16 munged sequence, modifies it according to the
+  string2key rules, and produces a UTF16 sequence
 
 The rules are:
 
-    1) convert any instance of 0xD800 - 0xDBFF (high surrogate)
+    1) any 0x0000 characters are mapped to 0x0001
+
+    2) convert any instance of 0xD800 - 0xDBFF (high surrogate)
        without an immediately following 0xDC00 - 0x0xDFFF (low surrogate) to
        U+FFFD (OBJECT REPLACEMENT CHARACTER).
 
-    2) the same for any low surrogate that was not preceded by a high surrogate.
+    3) the same for any low surrogate that was not preceded by a high surrogate.
+
  */
-static size_t utf8_munged_push(void *cd, const char **inbuf, size_t *inbytesleft,
+static size_t utf16_munged_pull(void *cd, const char **inbuf, size_t *inbytesleft,
 			       char **outbuf, size_t *outbytesleft)
 {
 	size_t in_left=*inbytesleft, out_left=*outbytesleft;
 	uint8_t *c = (uint8_t *)*outbuf;
 	const uint8_t *uc = (const uint8_t *)*inbuf;
 
-	while (in_left >= 2 && out_left >= 1) {
-		unsigned int codepoint;
+	while (in_left >= 2 && out_left >= 2) {
+		unsigned int codepoint = uc[0] | (uc[1]<<8);
 
-		if (uc[1] == 0 && !(uc[0] & 0x80)) {
-			/* simplest case */
-			c[0] = uc[0];
-			in_left  -= 2;
-			out_left -= 1;
-			uc += 2;
-			c  += 1;
-			continue;
+		if (codepoint == 0) {
+			codepoint = 1;
 		}
 
-		if ((uc[1]&0xf8) == 0) {
-			/* next simplest case */
-			if (out_left < 2) {
+		if ((codepoint & 0xfc00) == 0xd800) {
+			/* a high surrogate */
+			unsigned int codepoint2;
+			if (in_left < 4) {
+				codepoint = 0xfffd;
+				goto codepoint16;				
+			}
+			codepoint2 = uc[2] | (uc[3]<<8);
+			if ((codepoint2 & 0xfc00) != 0xdc00) {
+				/* high surrogate not followed by low
+				   surrogate: convert to 0xfffd */
+				codepoint = 0xfffd;
+				goto codepoint16;
+			}
+			if (out_left < 4) {
 				errno = E2BIG;
 				goto error;
 			}
-			c[0] = 0xc0 | (uc[0]>>6) | (uc[1]<<2);
-			c[1] = 0x80 | (uc[0] & 0x3f);
-			in_left  -= 2;
-			out_left -= 2;
-			uc += 2;
-			c  += 2;
+			memcpy(c, uc, 4);
+			in_left  -= 4;
+			out_left -= 4;
+			uc       += 4;
+			c        += 4;
 			continue;
 		}
 
-		if ((uc[1] & 0xfc) == 0xdc) {
-			/* low surrogate not preceded by high surrogate
-			   convert to 0xfffd */
-			codepoint = 0xfffd;
-			goto codepoint16;
-		}
-
-		if ((uc[1] & 0xfc) != 0xd8) {
-			codepoint = uc[0] | (uc[1]<<8);
-			goto codepoint16;
-		}
-
-		/* its the first part of a 4 byte sequence */
-		if (in_left < 4 || (uc[3] & 0xfc) != 0xdc) {
-			/* high surrogate not followed by low surrogate 
-			   convert to 0xfffd */
+		if ((codepoint & 0xfc00) == 0xdc00) {
+			/* low surrogate not preceded by high
+			   surrogate: convert to 0xfffd */
 			codepoint = 0xfffd;
-			goto codepoint16;
-		}
-
-		codepoint = 0x10000 + (uc[2] | ((uc[3] & 0x3)<<8) | 
-				       (uc[0]<<10) | ((uc[1] & 0x3)<<18));
-		
-		if (out_left < 4) {
-			errno = E2BIG;
-			goto error;
 		}
-		c[0] = 0xf0 | (codepoint >> 18);
-		c[1] = 0x80 | ((codepoint >> 12) & 0x3f);
-		c[2] = 0x80 | ((codepoint >> 6) & 0x3f);
-		c[3] = 0x80 | (codepoint & 0x3f);
-		
-		in_left  -= 4;
-		out_left -= 4;
-		uc       += 4;
-		c        += 4;
-		continue;
 
 	codepoint16:
-		if (out_left < 3) {
-			errno = E2BIG;
-			goto error;
-		}
-		c[0] = 0xe0 | (codepoint >> 12);
-		c[1] = 0x80 | ((codepoint >> 6) & 0x3f);
-		c[2] = 0x80 | (codepoint & 0x3f);
+		c[0] = codepoint & 0xFF;
+		c[1] = (codepoint>>8) & 0xFF;
 		
 		in_left  -= 2;
-		out_left -= 3;
+		out_left -= 2;
 		uc  += 2;
-		c   += 3;
+		c   += 2;
 		continue;		
 	}
 
diff --git a/lib/util/charset/tests/iconv.c b/lib/util/charset/tests/iconv.c
index 40e223b28f..1facea6136 100644
--- a/lib/util/charset/tests/iconv.c
+++ b/lib/util/charset/tests/iconv.c
@@ -398,10 +398,65 @@ static bool test_random_5m(struct torture_context *tctx)
 	return true;
 }
 
+
+static bool test_string2key(struct torture_context *tctx)
+{
+	uint16_t *buf;
+	char *dest = NULL;
+	TALLOC_CTX *mem_ctx = talloc_new(tctx);
+	ssize_t ret;
+	size_t len = (random()%1000)+1;
+	const uint16_t in1[10] = { 'a', 0xd805, 'b', 0xdcf0, 'c', 0, 'd', 'e', 'f', 'g' };
+	uint8_t le1[20];
+	uint8_t *munged1;
+	uint8_t *out1;
+	int i;
+	const char *correct = "a\357\277\275b\357\277\275c\001defg";
+
+	buf = talloc_size(mem_ctx, len*2);
+	generate_random_buffer((uint8_t *)buf, len*2);
+
+	torture_comment(tctx, "converting random buffer\n");
+
+	ret = convert_string_talloc(mem_ctx, CH_UTF16MUNGED, CH_UTF8, (void *)buf, len*2, (void**)&dest);
+	if (ret == -1) {
+		torture_fail(tctx, "Failed to convert random buffer\n");
+	}
+
+	for (i=0;i<10;i++) {
+		SSVAL(&le1[2*i], 0, in1[i]);
+	}
+
+	torture_comment(tctx, "converting fixed buffer to UTF16\n");
+
+	ret = convert_string_talloc(mem_ctx, CH_UTF16MUNGED, CH_UTF16, (void *)le1, 20, (void**)&munged1);
+	if (ret == -1) {
+		torture_fail(tctx, "Failed to convert fixed buffer to UTF16_MUNGED\n");
+	}
+
+	torture_assert(tctx, ret == 20, "conversion should give 20 bytes\n");
+
+	torture_comment(tctx, "converting fixed buffer to UTF8\n");
+
+	ret = convert_string_talloc(mem_ctx, CH_UTF16MUNGED, CH_UTF8, (void *)le1, 20, (void**)&out1);
+	if (ret == -1) {
+		torture_fail(tctx, "Failed to convert fixed buffer to UTF8\n");
+	}
+
+	torture_assert(tctx, strcmp(correct, out1) == 0, "conversion gave incorrect result\n");
+
+	talloc_free(mem_ctx);
+
+	return true;
+}
+
 struct torture_suite *torture_local_iconv(TALLOC_CTX *mem_ctx)
 {
 	struct torture_suite *suite = torture_suite_create(mem_ctx, "ICONV");
 
+	torture_suite_add_simple_test(suite, "string2key",
+				      test_string2key);
+
 	torture_suite_add_simple_test(suite, "next_codepoint()",
 				      test_next_codepoint);
 
@@ -410,6 +465,9 @@ struct torture_suite *torture_local_iconv(TALLOC_CTX *mem_ctx)
 
 	torture_suite_add_simple_test(suite, "5M random UTF-16LE sequences",
 				      test_random_5m);
+
+	torture_suite_add_simple_test(suite, "string2key",
+				      test_string2key);
 	return suite;
 }