libs/web: various changes in template library

- rename sanitize_pcdata() and sanitize_utf8() to pcdata() and utf8()
	- implement striptags()

libs/web/src/template_lualib.c

@@ -44,11 +44,11 @@ int template_L_parse(lua_State *L)
 	return rv;
 }
 
-int template_L_sanitize_utf8(lua_State *L)
+int template_L_utf8(lua_State *L)
 {
 	size_t len = 0;
 	const char *str = luaL_checklstring(L, 1, &len);
-	char *res = sanitize_utf8(str, len);
+	char *res = utf8(str, len);
 
 	if (res != NULL)
 	{
@@ -61,11 +61,28 @@ int template_L_sanitize_utf8(lua_State *L)
 	return 0;
 }
 
-int template_L_sanitize_pcdata(lua_State *L)
+int template_L_pcdata(lua_State *L)
 {
 	size_t len = 0;
 	const char *str = luaL_checklstring(L, 1, &len);
-	char *res = sanitize_pcdata(str, len);
+	char *res = pcdata(str, len);
+
+	if (res != NULL)
+	{
+		lua_pushstring(L, res);
+		free(res);
+
+		return 1;
+	}
+
+	return 0;
+}
+
+int template_L_striptags(lua_State *L)
+{
+	size_t len = 0;
+	const char *str = luaL_checklstring(L, 1, &len);
+	char *res = striptags(str, len);
 
 	if (res != NULL)
 	{
@@ -129,8 +146,9 @@ static int template_L_hash(lua_State *L) {
 /* module table */
 static const luaL_reg R[] = {
 	{ "parse",				template_L_parse },
-	{ "sanitize_utf8",		template_L_sanitize_utf8 },
+	{ "utf8",				template_L_utf8 },
-	{ "sanitize_pcdata",	template_L_sanitize_pcdata },
+	{ "pcdata",				template_L_pcdata },
+	{ "striptags",			template_L_striptags },
 	{ "load_catalog",		template_L_load_catalog },
 	{ "close_catalog",		template_L_close_catalog },
 	{ "change_catalog",		template_L_change_catalog },

libs/web/src/template_parser.c

@@ -244,7 +244,7 @@ template_format_chunk(struct template_parser *parser, size_t *sz)
 		switch (c->type)
 		{
 			case T_TYPE_TEXT:
-				escape_luastr(buf, c->s, c->e - c->s, 0);
+				luastr_escape(buf, c->s, c->e - c->s, 0);
 				break;
 
 			case T_TYPE_EXPR:
@@ -254,15 +254,15 @@ template_format_chunk(struct template_parser *parser, size_t *sz)
 				break;
 
 			case T_TYPE_INCLUDE:
-				escape_luastr(buf, c->s, c->e - c->s, 0);
+				luastr_escape(buf, c->s, c->e - c->s, 0);
 				break;
 
 			case T_TYPE_I18N:
-				translate_luastr(buf, c->s, c->e - c->s, 1);
+				luastr_translate(buf, c->s, c->e - c->s, 1);
 				break;
 
 			case T_TYPE_I18N_RAW:
-				translate_luastr(buf, c->s, c->e - c->s, 0);
+				luastr_translate(buf, c->s, c->e - c->s, 0);
 				break;
 
 			case T_TYPE_CODE:

libs/web/src/template_utils.c

@@ -278,7 +278,7 @@ static int _validate_utf8(unsigned char **s, int l, struct template_buffer *buf)
 }
 
 /* sanitize given string and replace all invalid UTF-8 sequences with "?" */
-char * sanitize_utf8(const char *s, unsigned int l)
+char * utf8(const char *s, unsigned int l)
 {
 	struct template_buffer *buf = buf_init(l);
 	unsigned char *ptr = (unsigned char *)s;
@@ -312,7 +312,7 @@ char * sanitize_utf8(const char *s, unsigned int l)
 /* Sanitize given string and strip all invalid XML bytes
  * Validate UTF-8 sequences
  * Escape XML control chars */
-char * sanitize_pcdata(const char *s, unsigned int l)
+char * pcdata(const char *s, unsigned int l)
 {
 	struct template_buffer *buf = buf_init(l);
 	unsigned char *ptr = (unsigned char *)s;
@@ -368,7 +368,67 @@ char * sanitize_pcdata(const char *s, unsigned int l)
 	return buf_destroy(buf);
 }
 
-void escape_luastr(struct template_buffer *out, const char *s, unsigned int l,
+char * striptags(const char *s, unsigned int l)
+{
+	struct template_buffer *buf = buf_init(l);
+	unsigned char *ptr = (unsigned char *)s;
+	unsigned char *end = ptr + l;
+	unsigned char *tag;
+	unsigned char prev;
+	char esq[8];
+	int esl;
+
+	for (prev = ' '; ptr < end; ptr++)
+	{
+		if ((*ptr == '<') && ((ptr + 2) < end) &&
+			((*(ptr + 1) == '/') || isalpha(*(ptr + 1))))
+		{
+			for (tag = ptr; tag < end; tag++)
+			{
+				if (*tag == '>')
+				{
+					if (!isspace(prev))
+						buf_putchar(buf, ' ');
+
+					ptr = tag;
+					prev = ' ';
+					break;
+				}
+			}
+		}
+		else if (isspace(*ptr))
+		{
+			if (!isspace(prev))
+				buf_putchar(buf, *ptr);
+
+			prev = *ptr;
+		}
+		else
+		{
+			switch(*ptr)
+			{
+				case '"':
+				case '\'':
+				case '<':
+				case '>':
+				case '&':
+					esl = snprintf(esq, sizeof(esq), "&#%i;", *ptr);
+					buf_append(buf, esq, esl);
+					break;
+
+				default:
+					buf_putchar(buf, *ptr);
+					break;
+			}
+
+			prev = *ptr;
+		}
+	}
+
+	return buf_destroy(buf);
+}
+
+void luastr_escape(struct template_buffer *out, const char *s, unsigned int l,
 				   int escape_xml)
 {
 	int esl;
@@ -411,7 +471,7 @@ void escape_luastr(struct template_buffer *out, const char *s, unsigned int l,
 	}
 }
 
-void translate_luastr(struct template_buffer *out, const char *s, unsigned int l,
+void luastr_translate(struct template_buffer *out, const char *s, unsigned int l,
 					  int escape_xml)
 {
 	char *tr;
@@ -420,11 +480,11 @@ void translate_luastr(struct template_buffer *out, const char *s, unsigned int l
 	switch (lmo_translate(s, l, &tr, &trlen))
 	{
 		case 0:
-			escape_luastr(out, tr, trlen, escape_xml);
+			luastr_escape(out, tr, trlen, escape_xml);
 			break;
 
 		case -1:
-			escape_luastr(out, s, l, escape_xml);
+			luastr_escape(out, s, l, escape_xml);
 			break;
 
 		default:

libs/web/src/template_utils.h

@@ -39,10 +39,11 @@ int buf_append(struct template_buffer *buf, const char *s, int len);
 int buf_length(struct template_buffer *buf);
 char * buf_destroy(struct template_buffer *buf);
 
-char * sanitize_utf8(const char *s, unsigned int l);
+char * utf8(const char *s, unsigned int l);
-char * sanitize_pcdata(const char *s, unsigned int l);
+char * pcdata(const char *s, unsigned int l);
+char * striptags(const char *s, unsigned int l);
 
-void escape_luastr(struct template_buffer *out, const char *s, unsigned int l, int escape_xml);
+void luastr_escape(struct template_buffer *out, const char *s, unsigned int l, int escape_xml);
-void translate_luastr(struct template_buffer *out, const char *s, unsigned int l, int escape_xml);
+void luastr_translate(struct template_buffer *out, const char *s, unsigned int l, int escape_xml);
 
 #endif