ox 1.6.8 → 1.6.9

data/README.md

@@ -34,13 +34,9 @@ A fast XML parser and Object marshaller as a Ruby gem.
 
 ## <a name="release">Release Notes</a>
 
-### Release 1.6.8
+### Release 1.6.9
 
- - Changed extconf.rb to use RUBY_PLATFORM.
-
-### Release 1.6.7
-
- - Now uses the encoding of the imput XML as the default encoding for the parsed output if the default options encoding is not set and the encoding is not set in the XML file prolog.
+ - Added support for BOM. They are honored for and handled correctly for UTF-8. Others cause encoding issues with Ruby or raise an error as others are not ASCII compatible..
 
 ## <a name="description">Description</a>
 

data/ext/ox/ox.c

@@ -131,8 +131,16 @@ static VALUE	xsd_date_sym;
 
 #if HAS_ENCODING_SUPPORT
 rb_encoding	*ox_utf8_encoding = 0;
+static rb_encoding	*ox_utf16le_encoding = 0;
+static rb_encoding	*ox_utf16be_encoding = 0;
+static rb_encoding	*ox_utf32le_encoding = 0;
+static rb_encoding	*ox_utf32be_encoding = 0;
 #else
 void		*ox_utf8_encoding = 0;
+static void	*ox_utf16le_encoding = 0;
+static void	*ox_utf16be_encoding = 0;
+static void	*ox_utf32le_encoding = 0;
+static void	*ox_utf32be_encoding = 0;
 #endif
 
 struct _Options	 ox_default_options = {
@@ -157,6 +165,55 @@ extern ParseCallbacks	ox_nomode_callbacks;
 
 static void	parse_dump_options(VALUE ropts, Options copts);
 
+static char*
+defuse_bom(char *xml, Options options) {
+    switch ((uint8_t)*xml) {
+    case 0xEF: /* UTF-8 */
+	if (0xBB == (uint8_t)xml[1] && 0xBF == (uint8_t)xml[2]) {
+	    options->rb_enc = ox_utf8_encoding;
+	    xml += 3;
+	} else {
+	    rb_raise(rb_eArgError, "Invalid BOM in XML string.\n");
+	}
+	break;
+#if 0
+    case 0xFE: /* UTF-16BE */
+	if (0xFF == (uint8_t)xml[1]) {
+	    options->rb_enc = ox_utf16be_encoding;
+	    xml += 2;
+	} else {
+	    rb_raise(rb_eArgError, "Invalid BOM in XML string.\n");
+	}
+	break;
+    case 0xFF: /* UTF-16LE or UTF-32LE */
+	if (0xFE == (uint8_t)xml[1]) {
+	    if (0x00 == (uint8_t)xml[2] && 0x00 == (uint8_t)xml[3]) {
+		options->rb_enc = ox_utf32le_encoding;
+		xml += 4;
+	    } else {
+		options->rb_enc = ox_utf16le_encoding;
+		xml += 2;
+	    }
+	} else {
+	    rb_raise(rb_eArgError, "Invalid BOM in XML string.\n");
+	}
+	break;
+    case 0x00: /* UTF-32BE */
+	if (0x00 == (uint8_t)xml[1] && 0xFE == (uint8_t)xml[2] && 0xFF == (uint8_t)xml[3]) {
+	    options->rb_enc = ox_utf32be_encoding;
+	    xml += 4;
+	} else {
+	    rb_raise(rb_eArgError, "Invalid BOM in XML string.\n");
+	}
+	break;
+#endif
+    default:
+	/* Let it fail if there is a BOM that is not UTF-8. Other BOM options are not ASCII compatible. */
+	break;
+    }
+    return xml;
+}
+
 /* call-seq: ox_default_options() => Hash
  *
  * Returns the default load and dump options as a Hash. The options are
@@ -312,9 +369,10 @@ set_def_opts(VALUE self, VALUE opts) {
  */
 static VALUE
 to_obj(VALUE self, VALUE ruby_xml) {
-    char	*xml;
-    size_t	len;
-    VALUE	obj;
+    char		*xml, *x;
+    size_t		len;
+    VALUE		obj;
+    struct _Options	options = ox_default_options;
 
     Check_Type(ruby_xml, T_STRING);
     /* the xml string gets modified so make a copy of it */
@@ -324,8 +382,9 @@ to_obj(VALUE self, VALUE ruby_xml) {
     } else {
 	xml = ALLOCA_N(char, len);
     }
-    strcpy(xml, StringValuePtr(ruby_xml));
-    obj = ox_parse(xml, ox_obj_callbacks, 0, &ox_default_options);
+    memcpy(xml, StringValuePtr(ruby_xml), len);
+    x = defuse_bom(xml, &options);
+    obj = ox_parse(x, ox_obj_callbacks, 0, &options);
     if (SMALL_XML < len) {
 	xfree(xml);
     }
@@ -341,9 +400,10 @@ to_obj(VALUE self, VALUE ruby_xml) {
  */
 static VALUE
 to_gen(VALUE self, VALUE ruby_xml) {
-    char		*xml;
+    char		*xml, *x;
     size_t		len;
     VALUE		obj;
+    struct _Options	options = ox_default_options;
 
     Check_Type(ruby_xml, T_STRING);
     /* the xml string gets modified so make a copy of it */
@@ -353,8 +413,9 @@ to_gen(VALUE self, VALUE ruby_xml) {
     } else {
 	xml = ALLOCA_N(char, len);
     }
-    strcpy(xml, StringValuePtr(ruby_xml));
-    obj = ox_parse(xml, ox_gen_callbacks, 0, &ox_default_options);
+    memcpy(xml, StringValuePtr(ruby_xml), len);
+    x = defuse_bom(xml, &options);
+    obj = ox_parse(x, ox_gen_callbacks, 0, &options);
     if (SMALL_XML < len) {
 	xfree(xml);
     }
@@ -413,6 +474,7 @@ load(char *xml, int argc, VALUE *argv, VALUE self, VALUE encoding) {
 	options.rb_enc = rb_enc_find(options.encoding);
     }
 #endif
+    xml = defuse_bom(xml, &options);
     switch (options.mode) {
     case ObjMode:
 	obj = ox_parse(xml, ox_obj_callbacks, 0, &options);
@@ -471,7 +533,7 @@ load_str(int argc, VALUE *argv, VALUE self) {
 #else
     encoding = Qnil;
 #endif
-    strcpy(xml, StringValuePtr(*argv));
+    memcpy(xml, StringValuePtr(*argv), len);
     obj = load(xml, argc - 1, argv + 1, self, encoding);
     if (SMALL_XML < len) {
 	xfree(xml);
@@ -804,6 +866,10 @@ void Init_ox() {
     rb_define_module_function(Ox, "cache8_test", cache8_test, 0);
 #if HAS_ENCODING_SUPPORT
     ox_utf8_encoding = rb_enc_find("UTF-8");
+    ox_utf16le_encoding = rb_enc_find("UTF-16LE");
+    ox_utf16be_encoding = rb_enc_find("UTF-16BE");
+    ox_utf32le_encoding = rb_enc_find("UTF-32LE");
+    ox_utf32be_encoding = rb_enc_find("UTF-32BE");
 #endif
 }
 

data/ext/ox/parse.c

@@ -745,26 +745,6 @@ ucs_to_utf8_chars(char *text, uint64_t u) {
     return text;
 }
 
-#if 0
-static char*
-uint64_to_chars(char *text, uint64_t u) {
-    int			reading = 0;
-    int			i;
-    unsigned char	c;
-
-    for (i = 56; 0 <= i; i -= 8) {
-	c = (unsigned char)((u >> i) & 0x00000000000000FFULL);
-	if (reading) {
-	    *text++ = (char)c;
-	} else if ('\0' != c) {
-	    *text++ = (char)c;
-	    reading = 1;
-	}
-    }
-    return text;
-}
-#endif
-
 static char*
 read_coded_chars(PInfo pi, char *text) {
     char	*b, buf[32];

data/lib/ox/version.rb

@@ -1,5 +1,5 @@
 
 module Ox
   # Current version of the module. 
-  VERSION = '1.6.8'
+  VERSION = '1.6.9'
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: ox
 version: !ruby/object:Gem::Version
-  version: 1.6.8
+  version: 1.6.9
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-11-18 00:00:00.000000000 Z
+date: 2012-11-25 00:00:00.000000000 Z
 dependencies: []
 description: ! "A fast XML parser and object serializer that uses only standard C
   lib.\n            \nOptimized XML (Ox), as the name implies was written to provide