Merge ascii and unicode handling in PLIST_STRING using UTF-8. Remove unicode related declaration in API (breaks API&ABI)

6 files changed, 46 insertions, 111 deletions

diff --git a/include/plist/plist.h b/include/plist/plist.h
index b7b0fa4..c289158 100644
--- a/include/plist/plist.h
+++ b/include/plist/plist.h
@@ -71,8 +71,6 @@ extern "C" {
 		PLIST_REAL,	/**< Real, scalar type */
 		PLIST_STRING,
 					/**< ASCII string, scalar type */
-		PLIST_UNICODE,
-					/**< Unicode strin, scalar type */
 		PLIST_ARRAY,/**< Ordered array, structured type */
 		PLIST_DICT,	/**< Unordered dictionary (key/value pair), structured type */
 		PLIST_DATE,	/**< Date, scalar type */
@@ -189,7 +187,7 @@ extern "C" {
  * (ie #PLIST_DICT or #PLIST_ARRAY).
  *
  * @param node the node to add a children to
- * @param val the string value encoded as an ASCII string (must be null terminated)
+ * @param val the string value encoded as an ASCII or UTF-8 string (must be null terminated)
  */
 	PLIST_API void plist_add_sub_string_el(plist_t node, const char *val);
 
@@ -231,15 +229,6 @@ extern "C" {
 	PLIST_API void plist_add_sub_data_el(plist_t node, const char *val, uint64_t length);
 
 /**
- * Add a subnode of type #PLIST_UNICODE to a node. The node must be of a structured type
- * (ie #PLIST_DICT or #PLIST_ARRAY).
- *
- * @param node the node to add a children to
- * @param val the unicode string encoded in UTF-8 (must be null terminated)
- */
-	PLIST_API void plist_add_sub_unicode_el(plist_t node, const char *val);
-
-/**
  * Add a subnode of type #PLIST_DATE to a node. The node must be of a structured type
  * (ie #PLIST_DICT or #PLIST_ARRAY).
  *
@@ -280,7 +269,7 @@ extern "C" {
  *
  * @param node the node
  * @param val a pointer to a C-string. This function allocates the memory,
- *            caller is responsible for freeing it.
+ *            caller is responsible for freeing it. Data is UTF-8 encoded.
  */
 	PLIST_API void plist_get_string_val(plist_t node, char **val);
 
@@ -322,16 +311,6 @@ extern "C" {
 	PLIST_API void plist_get_data_val(plist_t node, char **val, uint64_t * length);
 
 /**
- * Get the value of a #PLIST_UNICODE node.
- * This function does nothing if node is not of type #PLIST_UNICODE
- *
- * @param node the node
- * @param val a pointer to a C-string. This function allocates the memory,
- *            caller is responsible for freeing it. Data is UTF-8 encoded.
- */
-	PLIST_API void plist_get_unicode_val(plist_t node, char **val);
-
-/**
  * Get the value of a #PLIST_DATE node.
  * This function does nothing if node is not of type #PLIST_DATE
  *
diff --git a/src/bplist.c b/src/bplist.c
index 2a82b9f..6b2d2f3 100644
--- a/src/bplist.c
+++ b/src/bplist.c
@@ -24,6 +24,8 @@
 #include <stdio.h>
 #include <string.h>
 
+#include <libxml/tree.h>
+
 #include <plist/plist.h>
 #include "plist.h"
 
@@ -184,13 +186,28 @@ static plist_t parse_unicode_node(char *bnode, uint64_t size)
 {
 	plist_data_t data = plist_new_plist_data();
 	uint64_t i = 0;
-	data->type = PLIST_UNICODE;
-	data->unicodeval = (gunichar2 *) malloc(sizeof(gunichar2) * (size + 1));
-	memcpy(data->unicodeval, bnode, sizeof(gunichar2) * size);
-	data->unicodeval[sizeof(gunichar2) * size] = '\0';
-	data->length = size;
-	for (i = 0; i <= size; i++)
-		byte_convert((uint8_t*)(data->unicodeval + i), sizeof(gunichar2));
+	gunichar2 *unicodestr = NULL;
+	gchar *tmpstr = NULL;
+	int type = 0;
+	glong items_read = 0;
+	glong items_written = 0;
+	GError *error = NULL;
+
+	data->type = PLIST_STRING;
+	unicodestr = (gunichar2 *) malloc(sizeof(gunichar2) * size);
+	memcpy(unicodestr, bnode, sizeof(gunichar2) * size);
+	for (i = 0; i < size; i++)
+		byte_convert((uint8_t*)(unicodestr + i), sizeof(gunichar2));
+
+	tmpstr = g_utf16_to_utf8(unicodestr, size, &items_read, &items_written, &error);
+	free(unicodestr);
+
+	data->type = PLIST_STRING;
+	data->strval = (char *) malloc(sizeof(char) * (items_written + 1));
+	memcpy(data->strval, tmpstr, items_written);
+	data->strval[items_written] = '\0';
+	data->length = strlen(data->strval);
+	g_free(tmpstr);
 	return g_node_new(data);
 }
 
@@ -364,10 +381,6 @@ static gpointer copy_plist_data(gconstpointer src, gpointer data)
 	case PLIST_STRING:
 		dstdata->strval = strdup(srcdata->strval);
 		break;
-	case PLIST_UNICODE:
-		dstdata->unicodeval = (gunichar2 *) malloc(srcdata->length * sizeof(gunichar2));
-		memcpy(dstdata->unicodeval, srcdata->unicodeval, srcdata->length * sizeof(gunichar2));
-		break;
 	case PLIST_DATA:
 	case PLIST_ARRAY:
 		dstdata->buff = (uint8_t *) malloc(sizeof(uint8_t *) * srcdata->length);
@@ -520,10 +533,6 @@ static guint plist_data_hash(gconstpointer key)
 		buff = data->strval;
 		size = strlen(buff);
 		break;
-	case PLIST_UNICODE:
-		buff = (char *) data->unicodeval;
-		size = data->length;
-		break;
 	case PLIST_DATA:
 	case PLIST_ARRAY:
 	case PLIST_DICT:
@@ -752,6 +761,13 @@ void plist_to_bin(plist_t plist, char **plist_bin, uint32_t * length)
 	uint64_t *offsets = NULL;
 	uint8_t pad[6] = { 0, 0, 0, 0, 0, 0 };
 	uint8_t trailer[BPLIST_TRL_SIZE];
+	//for string
+	glong len = 0;
+	int type = 0;
+	glong items_read = 0;
+	glong items_written = 0;
+	GError *error = NULL;
+	gunichar2 *unicodestr = NULL;
 
 	//check for valid input
 	if (!plist || !plist_bin || *plist_bin || !length)
@@ -806,10 +822,16 @@ void plist_to_bin(plist_t plist, char **plist_bin, uint32_t * length)
 
 		case PLIST_KEY:
 		case PLIST_STRING:
-			write_string(bplist_buff, data->strval);
-			break;
-		case PLIST_UNICODE:
-			write_unicode(bplist_buff, data->unicodeval, data->length);
+			len = strlen(data->strval);
+			type = xmlDetectCharEncoding(data->strval, len);
+			if (XML_CHAR_ENCODING_UTF8 == type) {
+				unicodestr = g_utf8_to_utf16(data->strval, len, &items_read, &items_written, &error);
+				write_unicode(bplist_buff, unicodestr, items_written);
+				g_free(unicodestr);
+			}
+			else if (XML_CHAR_ENCODING_ASCII == type || XML_CHAR_ENCODING_NONE == type) {
+				write_string(bplist_buff, data->strval);
+			}
 			break;
 		case PLIST_DATA:
 			write_data(bplist_buff, data->buff, data->length);
diff --git a/src/plist.c b/src/plist.c
index 4fc2780..7949bce 100644
--- a/src/plist.c
+++ b/src/plist.c
@@ -53,9 +53,6 @@ static void plist_free_node(GNode * node, gpointer none)
 		case PLIST_STRING:
 			free(data->strval);
 			break;
-		case PLIST_UNICODE:
-			free(data->unicodeval);
-			break;
 		case PLIST_DATA:
 			free(data->buff);
 			break;
@@ -90,10 +87,6 @@ static plist_t plist_add_sub_element(plist_t node, plist_type type, const void *
 		//only structured types are allowed to have nulll value
 		if (value || (!value && (type == PLIST_DICT || type == PLIST_ARRAY))) {
 
-			glong len = 0;
-			glong items_read = 0;
-			glong items_written = 0;
-			GError *error = NULL;
 			plist_t subnode = NULL;
 
 			//now handle value
@@ -115,11 +108,6 @@ static plist_t plist_add_sub_element(plist_t node, plist_type type, const void *
 			case PLIST_STRING:
 				data->strval = strdup((char *) value);
 				break;
-			case PLIST_UNICODE:
-				len = strlen((char *) value);
-				data->unicodeval = g_utf8_to_utf16((char *) value, len, &items_read, &items_written, &error);
-				data->length = items_written;
-				break;
 			case PLIST_DATA:
 				data->buff = (uint8_t*)malloc(length);
 				memcpy(data->buff, value, length);
@@ -210,9 +198,6 @@ static char compare_node_value(plist_type type, plist_data_t data, const void *v
 	case PLIST_STRING:
 		res = !strcmp(data->strval, ((char *) value));
 		break;
-	case PLIST_UNICODE:
-		res = !memcpy(data->unicodeval, value, length);
-		break;
 	case PLIST_DATA:
 		res = !memcmp(data->buff, (char *) value, length);
 		break;
@@ -262,11 +247,6 @@ plist_t plist_find_node_by_string(plist_t plist, const char *value)
 
 static void plist_get_type_and_value(plist_t node, plist_type * type, void *value, uint64_t * length)
 {
-	//for unicode
-	glong len = 0;
-	glong items_read = 0;
-	glong items_written = 0;
-	GError *error = NULL;
 	plist_data_t data = NULL;
 
 	if (!node)
@@ -291,10 +271,6 @@ static void plist_get_type_and_value(plist_t node, plist_type * type, void *valu
 	case PLIST_STRING:
 		*((char **) value) = strdup(data->strval);
 		break;
-	case PLIST_UNICODE:
-		len = data->length;
-		*((char **) value) = g_utf16_to_utf8(data->unicodeval, len, &items_read, &items_written, &error);
-		break;
 	case PLIST_DATA:
 		*((uint8_t **) value) = (uint8_t *) malloc(*length * sizeof(uint8_t));
 		memcpy(*((uint8_t **) value), data->buff, *length * sizeof(uint8_t));
@@ -360,11 +336,6 @@ void plist_add_sub_data_el(plist_t node, const char *val, uint64_t length)
 	plist_add_sub_element(node, PLIST_DATA, val, length);
 }
 
-void plist_add_sub_unicode_el(plist_t node, const char *val)
-{
-	plist_add_sub_element(node, PLIST_UNICODE, val, strlen(val));
-}
-
 void plist_add_sub_date_el(plist_t node, int32_t sec, int32_t usec)
 {
 	GTimeVal val = { sec, usec };
@@ -423,15 +394,6 @@ void plist_get_data_val(plist_t node, char **val, uint64_t * length)
 		plist_get_type_and_value(node, &type, (void *) val, length);
 }
 
-void plist_get_unicode_val(plist_t node, char **val)
-{
-	plist_type type = plist_get_node_type(node);
-	uint64_t length = 0;
-	if (PLIST_UNICODE == type)
-		plist_get_type_and_value(node, &type, (void *) val, &length);
-	assert(length == strlen(*val));
-}
-
 void plist_get_date_val(plist_t node, int32_t * sec, int32_t * usec)
 {
 	plist_type type = plist_get_node_type(node);
@@ -476,11 +438,6 @@ gboolean plist_data_compare(gconstpointer a, gconstpointer b)
 			return TRUE;
 		else
 			return FALSE;
-	case PLIST_UNICODE:
-		if (!memcmp(val_a->unicodeval, val_b->unicodeval, val_a->length))
-			return TRUE;
-		else
-			return FALSE;
 
 	case PLIST_DATA:
 		if (!memcmp(val_a->buff, val_b->buff, val_a->length))
diff --git a/src/plist.h b/src/plist.h
index ebbb0b5..723ed87 100644
--- a/src/plist.h
+++ b/src/plist.h
@@ -36,7 +36,6 @@ struct plist_data_s {
 		uint64_t intval;
 		double realval;
 		char *strval;
-		gunichar2 *unicodeval;
 		uint8_t *buff;
 		GTimeVal timeval;
 	};
diff --git a/src/xplist.c b/src/xplist.c
index 2bee7c7..38cc4fe 100644
--- a/src/xplist.c
+++ b/src/xplist.c
@@ -126,12 +126,6 @@ static void node_to_xml(GNode * node, gpointer xml_struct)
 	//for base64
 	gchar *valtmp = NULL;
 
-	//for unicode
-	glong len = 0;
-	glong items_read = 0;
-	glong items_written = 0;
-	GError *error = NULL;
-
 	uint32_t i = 0;
 
 	if (!node)
@@ -165,12 +159,6 @@ static void node_to_xml(GNode * node, gpointer xml_struct)
 		val = g_strdup(node_data->strval);
 		break;
 
-	case PLIST_UNICODE:
-		tag = XPLIST_STRING;
-		len = node_data->length;
-		val = g_utf16_to_utf8(node_data->unicodeval, len, &items_read, &items_written, &error);
-		break;
-
 	case PLIST_KEY:
 		tag = XPLIST_KEY;
 		val = g_strdup((gchar *) node_data->strval);
@@ -233,11 +221,7 @@ static void xml_to_node(xmlNodePtr xml_node, plist_t * plist_node)
 	plist_t subnode = NULL;
 
 	//for string
-	unsigned char *tmp = NULL;
 	glong len = 0;
-	glong items_read = 0;
-	glong items_written = 0;
-	GError *error = NULL;
 	int type = 0;
 
 	if (!xml_node)
@@ -301,16 +285,11 @@ static void xml_to_node(xmlNodePtr xml_node, plist_t * plist_node)
 		if (!xmlStrcmp(node->name, XPLIST_STRING)) {
 			xmlChar *strval = xmlNodeGetContent(node);
 			len = strlen((char *) strval);
-			items_read = 0;
-			items_written = 0;
-			error = NULL;
 			type = xmlDetectCharEncoding(strval, len);
 
-			if (XML_CHAR_ENCODING_UTF8 == type) {
-				data->unicodeval = g_utf8_to_utf16((char *) strval, len, &items_read, &items_written, &error);
-				data->type = PLIST_UNICODE;
-				data->length = items_written;
-			} else if (XML_CHAR_ENCODING_ASCII == type || XML_CHAR_ENCODING_NONE == type) {
+			if (XML_CHAR_ENCODING_UTF8 == type ||
+				XML_CHAR_ENCODING_ASCII == type ||
+				XML_CHAR_ENCODING_NONE == type) {
 				data->strval = strdup((char *) strval);
 				data->type = PLIST_STRING;
 				data->length = strlen(data->strval);
diff --git a/swig/plist.i b/swig/plist.i
index 79298e6..b6f9f78 100644
--- a/swig/plist.i
+++ b/swig/plist.i
@@ -30,7 +30,6 @@ typedef enum {
 	PLIST_UINT,
 	PLIST_REAL,
 	PLIST_STRING,
-	PLIST_UNICODE,
 	PLIST_ARRAY,
 	PLIST_DICT,
 	PLIST_DATE,